+++ /dev/null
-# ChangeLog for Portage; the Gentoo Linux ports system
-# Copyright 1999-2005 Gentoo Foundation; Distributed under the GPL v2
-# $Id: ChangeLog,v 1.796.2.146 2005/08/20 17:24:30 jstubbs Exp $
-
- MAJOR CHANGES in 2.0.51:
- 1. /var/cache/edb/virtuals is no longer used at all. It's calculated now.
- 2. /var/cache/edb/world is now /var/lib/portage/world.
- 3. /etc/portage/profile/virtuals is _USER_ configs only.
-
- 01 Oct 2005; Brian Harring <ferringb@gentoo.org> pym/portage.py:
- selinux_enabled tweak, instead of relying on exception thrown on import,
- check the actual enabled value (forward compatibility).
-
- 01 Oct 2005; Brian Harring <ferringb@gentoo.org> bin/repoman:
- Touch ups to file.name check.
-
- 01 Oct 2005; Brian Harring <ferringb@gentoo.org> bin/ebuild,
- pym/portage.py:
- Fixed ebuild some-ebuild setup unpack so it functions like
- ebuild some-ebuild unpack does (execute just unpack due to
- env issues, see comments in ebuild clarifying if curious).
-
- 28 Sep 2005; Brian Harring <ferringb@gentoo.org> bin/repoman:
- Fixup of the file.name check so it doesn't catch CVS dirs.
-
- 27 Sep 2005; Brian Harring <ferringb@gentoo.org> bin/ebuild.sh:
- No more portageq in the global scope.
- Instant die.
-
- 27 Sep 2005; Brian Harring <ferringb@gentoo.org> bin/ebuild.sh,
- pym/portage.py:
- Tweak to make EAPI work for existing vdb entries.
-
- 24 Sep 2005; Brian Harring <ferringb@gentoo.org> pym/portage.py,
- bin/emerge:
- Final emerge metadata/cache cleanups; emerge --metadata no longer
- requires root (we run searches as non root, which can update the cache,
- so emerge --metadata as root is daft), and silenced auxdb exceptions
- when doing metadata transfers- preparation for when the default backend
- is changed. If that weren't in, user would see > 20,0000 lines of
- pointless cruft output.
-
- 24 Sep 2005; Brian Harring <ferringb@gentoo.org> pym/portage.py,
- +pym/portage_db_metadata.py, +pym/portage_db_flat_hash.py,
- pym/portage_db_flat.py, pym/portage_const.py:
- EAPI awareness, and support for flat_hash metadata/cache.
- Later autodetects if the rsync cache is flat_hash, or flat_list,
- and interprets accordingly.
-
- Flat_hash is intended to make it such that rsync cache breakages
- for older portage versions no longer occur; the container format
- can hold pretty much anything across EAPI versions.
- Additionally... EAPI is intended such that additions to the ebuild
- env/helpers don't puke if you're running to old of a portage version;
- it allows us to 'version' the ebuild env/funcs, and do masking dependant
- on that versioning.
-
- 23 Sep 2005; Brian Harring <ferringb@gentoo.org> bin/emerge:
- cleanse_cache cleanups, should kill off all remaining issues
- in bug #100444 .
-
- 23 Sep 2005; Brian Harring <ferringb@gentoo.org> bin/repoman:
- bug #106544, glep31 enforcement checks.
-
- 20 Sep 2005; Brian Harring <ferringb@gentoo.org> pym/portage.py:
- Cleanup after python, remove module from sys.modules that was a failed import
- for selinux code- identified by Tudor Alexandru Dragos in bug #106363 .
-
- 18 Sep 2005; Mike Frysinger <vapier@gentoo.org> bin/etc-update:
- Make sure diff works before checking config files #48184.
-
- 08 Sep 2005; Brian Harring <ferringb@gentoo.org> bin/ebuild.sh: Fixed
- unpack so that it actually works now. ${tarvars} placement was a bit
- wrong, which is odd considering it shouldn't have changed.
-
- 08 Sep 2005; Jason Stubbs <jstubbs@gentoo.org> bin/emerge: Only warn on
- removal of system virtuals if there is only one satisfying package.
-
- 21 Aug 2005; Jason Stubbs <jstubbs@gentoo.org> bin/ebuild.sh: Skip protecting
- ownership of symlinks. Patch by truedfx #94199
-
- 14 Aug 2005; Mike Frysinger <vapier@gentoo.org> bin/prepstrip:
- Don't try to strip if `file` failed #102499 by Daniel Drake.
-
- 10 Aug 2005; Mike Frysinger <vapier@gentoo.org> bin/quickpkg:
- Call gawk instead of awk so non-GNU systems work #102050 by Stephen Bennett.
- Also clean up the script a bit.
-
- 08 Aug 2005; Brian Harring <ferringb@gentoo.org> bin/ebuild.sh:
- fixed bug #101035 , basically checking pipestatus during unpack for
- piped unpacks.
-
- 07 Aug 2005; Mike Frysinger <vapier@gentoo.org> bin/ebuild.sh:
- Add support for unpacking rar archives.
-
- 04 Aug 2005; Brian Harring <ferringb@gentoo.org> bin/emerge:
- bug 64585, sanity check on cache cleansing.
-
- 04 Aug 2005: Brian Harring <ferringb@gentoo.org> bin/emerge:
- It's not perfect, but a fix for bug #96410. A better fix would be
- to get the categories limiter out of settings...
-
- 29 Jul 2005; Mike Frysinger <vapier@gentoo.org> bin/doman:
- Back port from portage HEAD:
- Remove old unused options (-x11/-gnome/-kde/etc...) and cleanup.
-
- 04 Jul 2005; Mike Frysinger <vapier@gentoo.org>
- bin/ebuild-default-functions.sh:
- Fix gnuconfig find expression #93363 by TGL.
-
- 23 Jun 2005; Mike Frysinger <vapier@gentoo.org>
- bin/ebuild-default-functions.sh:
- Update the scanelf RUNPATH check to abort on null paths.
-
- 17 Jun 2005; Mike Frysinger <vapier@gentoo.org> bin/repoman:
- Report exit status if gpg failed to return with 0.
-
- 14 Jun 2005; Marius Mauch <genone@gentoo.org> man/portage.5:
- Remove references to non-existant $PORTDIR/profiles/use.mask and revert
- the "fix" for bug 87173.
-
- 13 Jun 2005; Mike Frysinger <vapier@gentoo.org> bin/ebuild.sh:
- Use the %p output modifier instead of %F so we don't have to update the
- output by removing all the $D references.
-
- 29 May 2005; Jason Stubbs <jstubbs@gentoo.org> bin/ebuild.sh: Made
- /etc/profile.env sourced regardless of USERLAND. Made copying of config.sub
- and config.guess --force'd as they are sometimes u-w. Replaced a if/else of
- chgrp root/wheel with chgrp 0.
-
- 29 May 2005; Jason Stubbs <jstubbs@gentoo.org> bin/emerge: Moved the system
- package unmerge check after slots are checked and removed the countdown for
- when --ask is specified. Added a chdir to the removal of old cache call so
- that there are no errors due to userpriv and being in a non-portage-readable
- directory.
-
- 29 May 2005; Jason Stubbs <jstubbs@gentoo.org> bin/etc-update: Moved the
- temporary work dir from /tmp/$$ to ${PORTAGE_TMPDIR}/$$.
-
- 29 May 2005; Jason Stubbs <jstubbs@gentoo.org> bin/prepinfo bin/prepman:
- Reordered the -{min,max}depth and -name arguments to find so that warnings
- are not produced.
-
- 29 May 2005; Jason Stubbs <jstubbs@gentoo.org> pym/portage.py: Added support
- for FreeBSD's ldconfig. Fixed an incorrect chown of BUILD_PREFIX instead of
- PORT_LOGDIR.
-
- 29 May 2005; Mike Frysinger <vapier@gentoo.org> bin/ebuild.sh bin/prepstrip:
- Backport scanelf QA checks from HEAD.
-
- 25 May 2005; Brian Harring <ferringb@gentoo.org> bin/ebuild.sh: Bug 93293,
- die + stat_perms issues. Tweaked do_stat definition so it's one time instead
- of ongoing also.
-
- 19 May 2005; Jason Stubbs <jstubbs@gentoo.org> pym/portage.py: Disabled
- userpriv as a user globally instead of only in fetch.
-
- 19 May 2005; Jason Stubbs <jstubbs@gentoo.org> bin/emake: Made the
- make command configurable but defaulted to "make".
-
- 19 May 2005; Jason Stubbs <jstubbs@gentoo.org> bin/emerge: Fixed an
- issue that would cause binary package reinstalls when IUSE disagreed.
-
- 19 May 2005; Jason Stubbs <jstubbs@gentoo.org> bin/repoman: Fixed an
- issue where the portage version wasn't appended to the commit message
- when specified on the command line. Fixed the detection of files above
- the package level during Manifest generation.
-
- 19 May 2005; Jason Stubbs <jstubbs@gentoo.org> pym/portage.py: Renamed
- the userpriv FEATURE to userfetch for the fetch() phase and fixed a bug
- where privileges would try to be dropped even as non-root. Fixed a bug
- where files were attempted to be fetched when only a manifest recreation
- is requested.
-
- 17 May 2005; Jason Stubbs <jstubbs@gentoo.org> pym/portage.py: Stopped
- attempting to fetch files when only the manifest is being generated.
-
-*portage-2.0.51.22 (15 May 2005): Maintainence Release
-
- 15 May 2005; Jason Stubbs <jstubbs@gentoo.org> bin/dodoc: Removed the dodoc
- new testing features from head's version of dodoc.
-
- 15 May 2005; Jason Stubbs <jstubbs@gentoo.org> bin/emerge: Made --newuse
- take priority over --usepkg.
-
- 15 May 2005; Jason Stubbs <jstubbs@gentoo.org> bin/repoman: Added HOMEPAGE
- to the list of vars that must be defined. Restored detection of the
- digest.partial check. Made a commit message required.
-
- 15 May 2005; Jason Stubbs <jstubbs@gentoo.org> pym/portage.py: Fixed the
- definition of PYTHONPATH in env_update(). Fixed the handling of quotes in
- env_update().
-
- 13 May 2005; Jason Stubbs <jstubbs@gentoo.org> cnf/dispatch-conf.conf
- bin/dispatch-conf: Removed the pager config option and made if part of the
- diff command option.
-
- 13 May 2005; Jason Stubbs <jstubbs@gentoo.org> bin/dodoc: Backported vapier's
- fix for 0-byte files causing a spurious "file not found" error.
-
- 13 May 2005; Jason Stubbs <jstubbs@gentoo.org> bin/emerge: Remove --verbose
- and --tree from options when using --resume as the combination is currently
- broken.
-
- 13 May 2005; Jason Stubbs <jstubbs@gentoo.org> bin/repoman: Added three new
- checks on ebuild's PROVIDEs. Optimized out a lot of unnecessary aux_get()
- calls. Modified to commit an unsigned Manifest before committing the signed
- one to help ensure broken digests don't make it to rsync.
-
- 13 May 2005; Jason Stubbs <jstubbs@gentoo.org> pym/portage.py: Added SCCS to
- the list of directories to ignore when ignorecvs is set in cacheddir. Drop
- priveleges during fetch when userpriv is set. Added message display when
- changing permissions on ccache dirs.
-
- 06 May 2005; Brian Harring <ferringb@gentoo.org> bin/ebuild: fix portage_util
- import on osx.
-
- 05 May 2005; Jason Stubbs <jstubbs@gentoo.org> bin/etc-update: Fixed the
- regex and added locale overrides to the use of cut as per bug #91159.
-
- 05 May 2005; Jason Stubbs <jstubbs@gentoo.org> cnf/make.globals: Made
- CONFIG_PROTECT default to /etc only as packages are augmenting it via env.d
- where necessary.
-
- 05 May 2005; Jason Stubbs <jstubbs@gentoo.org> bin/emerge: Reverted deletion
- of apparently unused code as it was being used in the case of --noreplace.
-
- 04 May 2005; Marius Mauch <genone@gentoo.org> bin/g-cpan.pl, man/g-cpan.pl.1:
- removed g-cpan.pl as it's now a standalone tool.
-
- 02 May 2005; Jason Stubbs <jstubbs@gentoo.org> pym/portage.py: Added back
- support for ~* and * in package.keywords as it got dropped at some point.
-
-*portage-2.0.51.21 (01 May 2005): Maintainence Release
-
- 01 May 2005; Jason Stubbs <jstubbs@gentoo.org> bin/ebuild.sh: Fixed an
- inverse logic bug in the setting of ccache size.
-
- 01 May 2005; Jason Stubbs <jstubbs@gentoo.org>: Removed g-cpan.pl as it
- is now maintained externally. Removed other old and/or unmaintained
- scripts.
-
- 01 May 2005; Jason Stubbs <jstubbs@gentoo.org> pym/portage.py: Dropped
- confmem from 16 previous files down to 1 previous file so that upgrading
- and downgrading behaves in terms of config files.
-
- 01 May 2005; Jason Stubbs <jstubbs@gentoo.org> pym/portage.py: Added
- checking of overlays for package.mask, categories and others.
-
- 01 May 2005; Jason Stubbs <jstubbs@gentoo.org> bin/ebuild.sh: Skip testing
- if has already been performed (indicated by .tested)
-
- 01 May 2005; Jason Stubbs <jstubbs@gentoo.org> pym/portage.py: Removed
- debugging information and added a fix for merging a symlink to a
- directory over a file.
-
- 01 May 2005; Jason Stubbs <jstubbs@gentoo.org> cnf/make.globals: Added
- a basic CONFIG_PROTECT setting for those users who use only binary
- packages and don't have a portage tree.
-
- 29 Apr 2005; Jason Stubbs <jstubbs@gentoo.org> bin/doman: Added Mike's
- patch to broaden the accepted man pages from [1-8n] to [0-9n](|f|p|pm)
-
- 29 Apr 2005; Jason Stubbs <jstubbs@gentoo.org> bin/emerge pym/portage.py:
- Added an implementation for the emerge config action and a supporting
- method findname on the vardbapi class.
-
- 29 Apr 2005; Jason Stubbs <jstubbs@gentoo.org> man/portage.5: Removed
- invalid example from portage.keywords.
-
- 29 Apr 2005; Jason Stubbs <jstubbs@gentoo.org> bin/emerge: Drop privs
- when doing cache cleansing after syncing.
-
- 29 Apr 2005; Jason Stubbs <jstubbs@gentoo.org> pym/portage.py: Fixed
- several issues with userpriv.
-
- 29 Apr 2005; Jason Stubbs <jstubbs@gentoo.org> bin/ebuild.sh: Made the
- autoconfig feature non-optional.
-
- 29 Apr 2005; Jason Stubbs <jstubbs@gentoo.org> pym/portage.py: Removed
- running of depscan.sh from env-update.
-
- 29 Apr 2005; Jason Stubbs <jstubbs@gentoo.org> pym/portage.py: Applied
- solar's patches from bug #90720 that allow most ebuild operations to work
- as any user in the portage group.
-
- 29 Apr 2005; Jason Stubbs <jstubbs@gentoo.org> pym/portage.py: Fixed a bug
- that would cause an empty CBUILD to be defined if CHOST is not defined.
-
- 29 Apr 2005; Jason Stubbs <jstubbs@gentoo.org> cnf/make.globals.*: Removed
- the now deprecated arch-specific make.globals files.
-
- 29 Apr 2005; Jason Stubbs <jstubbs@gentoo.org> bin/dispatch-conf
- pym/dispatch-conf.py: Fixed a typo with the added merge option. Added
- hard-coded default merge option so that dispatch-conf can be used to do
- a line by line of dispatch-conf.conf.
-
- 29 Apr 2005; Jason Stubbs <jstubbs@gentoo.org> bin/ebuild.sh: Updated econf
- to use LOCAL_EXTRA_ECONF rather than EXTRA_ECONF directly, as noted by
- eradicator. Removed the forced resizing of ccache when CCACHE_SIZE is
- undefined.
-
- 29 Apr 2005; Jason Stubbs <jstubbs@gentoo.org> bin/emerge: Fixed bug in
- --info during sorting when two versions have only differing -r component.
- Ported Brian's regen method to --metadata for faster updates.
-
- 29 Apr 2005; Jason Stubbs <jstubbs@gentoo.org> bin/repoman: Fixed an
- off-by-one indexing into a list during scanning for Manfiest commits.
-
- 29 Apr 2005; Jason Stubbs <jstubbs@gentoo.org> pym/portage.py: Fixed a bug
- in the checking of whether sandbox is usable when usersandbox is enabled.
- Combined u+w and g+w mods on ccache dir into the one exec. Added missing g+s
- mods on ccache dir, which was causing mods at each check. Moved mtime check
- in aux_get so that it applies to updating from the metadata cache as well.
- Relocated the fix for mynewcat outside of the loop to prevent needless
- re-calculation. Added debugging code for bug #71787.
-
- 24 Apr 2005; Masatomo Nakano <nakano@gentoo.org> pym/portage.py
- Fixed undefined value causing traceback when moving package in binarytree.
-
-*portage-2.0.51.20 (23 Apr 2005): Maintainence Release
-
- 23 Apr 2005; Jason Stubbs <jstubbs@gentoo.org> pym/portage_const.py:
- Adjusted sandbox path to match the external sandbox.
-
- 23 Apr 2005; Jason Stubbs <jstubbs@gentoo.org> bin/repoman: Added a 3 second
- delay after a failed Manifest commit. Fixed bug where files at the category
- level were being treated as packages.
-
- 23 Apr 2005; Jason Stubbs <jstubbs@gentoo.org> pym/portage.py portage_util.py
- portage_db_cpickle.py: Replaced HIGHEST_PROTOCOL with it's equivalent -1 to
- ensure python-2.2 compatibility. Added back ChangeLog and metadata.xml to
- Manifest generation for portage compatibility.
-
- 21 Apr 2005; Jason Stubbs <jstubbs@gentoo.org> pym/portage.py: Adding
- setting of CBUILD to CHOST if it is not set. Fixed getmaskingstatus to
- choose an arch-based keyword over -*.
-
- 21 Apr 2005; Jason Stubbs <jstubbs@gentoo.org> bin/ebuild.sh: Removed
- stale code from dyn_setup.
-
- 19 Apr 2005; Brian Harring <ferringb@gentoo.org> pym/portage_db_flat.py:
- Correction for string method name, again, danke to swegener.
-
- 19 Apr 2005; Jason Stubbs <jstubbs@gentoo.org> bin/ebuild: Added a missing
- import of portage_util.
-
- 19 Apr 2005; Jason Stubbs <jstubbs@gentoo.org> pym/portage.py: Fixed my
- assumption that ChangeLog and metadata.xml will exist during digest creation.
-
- 17 Apr 2005; Brian Harring <ferringb@gentoo.org> pym/portage.py: eclass cache
- touche up, mtime wasn't being properly checked (thanks to swegener for catching
- it).
-
- 17 Apr 2005; Jason Stubbs <jstubbs@gentoo.org> pym/portage_exec.py
- pym/portage.py: Added checks to see if the sandbox is executable and disabled
- when it isn't.
-
- 17 Apr 2005; Jason Stubbs <jstubbs@gentoo.org> bin/dispatch-conf
- cnf/dispatch-conf.conf pym/dispatch-conf.py: Made the merge command
- configurable. Fixed a bug that caused conflicts to make their way into the
- final configuration file. Made the config archive dir automatically created
- if there is nothing in the way.
-
- 17 Apr 2005; Jason Stubbs <jstubbs@gentoo.org> pym/output.py: Reverted the
- feature to make use of screen titles.
-
- 17 Apr 2005; Jason Stubbs <jstubbs@gentoo.org> bin/emerge: Made detection
- of --ask capability use stdin instead of stdout.
-
- 15 Apr 2005; Jason Stubbs <jstubbs@gentoo.org> pym/portage.py: Removed
- ChangeLog and metadata.xml from Manifest creation and checking. Quietened
- failed mtimedb writes. Added patch to allow fetching to read-only DISTDIR.
- Added logic to detect changing between userpriv and non-userpriv and
- update the ccache permissions at each change.
-
- 15 Apr 2005; Jason Stubbs <jstubbs@gentoo.org> cnf/make.globals*: Removed
- vars not directly related to portage. Architecture specific ones are going
- into the relevant profiles. USE_ORDER and CONFIG_PROTECT* will likely be
- going into base.
-
- 14 Apr 2005; Jason Stubbs <jstubbs@gentoo.org> pym/portage.py: Applied
- patch from #69763 to allow fetching to a read-only distfiles dir.
-
- 13 Apr 2005; Jason Stubbs <jstubbs@gentoo.org> bin/repoman: Downgraded
- the nesteddie check to a warning as it sometimes has false positives.
-
- 13 Apr 2005; Jason Stubbs <jstubbs@gentoo.org> cnf/make.conf*: Removed
- size estimation of PORTDIR.
-
- 13 Apr 2005; Jason Stubbs <jstubbs@gentoo.org> pym/portage.py: Added
- verification of package.provided contents. Fixed a bug with an incorrect
- error on binary package category moves. Fixed a small bug that would
- cause a crash on an invalid PRELINK_PATH_MASK. Added signal handling of
- SIGPIPE. Added deep fixing of permissions on CCACHE_DIR when changing
- between userpriv states.
-
- 13 Apr 2005; Jason Stubbs <jstubbs@gentoo.org> bin/emerge: Added sorting to
- versions listed in emerge info output.
-
- 13 Apr 2005; Jason Stubbs <jstubbs@gentoo.org> man/emerge.1: Removed --inject
- documentation.
-
- 12 Apr 2005; Jason Stubbs <jstubbs@gentoo.org> bin/emerge: Added --nocolor
- option courtesy of Jason Cooper on gentoo-user@g.o.
-
- 12 Apr 2005; Jason Stubbs <jstubbs@gentoo.org> bin/repoman: Fixed the date
- parsing of repoman to allow ranged and single years beginning 1999. #85242
-
- 12 Apr 2005; Jason Stubbs <jstubbs@gentoo.org> pym/portage.py: Added
- pusedict optimization. #85786
-
- 12 Apr 2005; Jason Stubbs <jstubbs@gentoo.org> bin/emerge: Fixed emerge to
- always add packages to world unless --oneshot is specified.
-
- 12 Apr 2005; Jason Stubbs <jstubbs@gentoo.org> bin/dispatch-conf:
- Replaced shutil.movefile() calls for python-2.2 compatibility.
-
- 12 Apr 2005; Jason Stubbs <jstubbs@gentoo.org> bin/portageq pym/portage.py:
- Utilized PORTAGE_CALLER to only print profile deprecation noticies when
- emerge first starts up.
-
- 12 Apr 2005; Jason Stubbs <jstubbs@gentoo.org> bin/repoman:
- Limited digest.assumed and ebuild.allmasked (by default) to scans at
- the package level only. Removed IUSE from required variables. Fixed
- duplicate reporting of required variables. Adjusted size limit of
- files/* to 20480 bytes. Fixed allmasked check. Removed reinitialization
- of each profile at every usage.
-
- 12 Apr 2005; Jason Stubbs <jstubbs@gentoo.org> pym/portage.py: Fixed
- spelling error. Removed code from portdbapi.gvisible that was only
- utilized by repoman's (broken) allmasked check.
-
- 03 Apr 2005; Jason Stubbs <jstubbs@gentoo.org> bin/emerge: Modified to
- ignore blockers when using --buildpkgonly.
-
- 02 Apr 2005; Jason Stubbs <jstubbs@gentoo.org> pym/portage.py:
- Made autouse calculated only at import time to ensure that use flags
- from dep calculation are the same as those used at build time.
-
- 02 Apr 2005; Jason Stubbs <jstubbs@gentoo.org> pym/portage_dep.py:
- Fixed a bug where a return type was a tuple instead of a list in
- paren_reduce().
-
- 23 Mar 2005; Jason Stubbs <jstubbs@gentoo.org> bin/dispatch-conf:
- Removed the -a argument from all diff usage.
-
- 08 Mar 2005; Brian Harring <ferringb@gentoo.org> pym/portage.py:
- Tweak to ensure restrict is dumped to the env, with PORTAGE_RESTRICT
- being dumped also.
-
- 06 Mar 2005; Jason Stubbs <jstubbs@gentoo.org> bin/emerge: Added a check
- to make sure the file exists before working with it, when called with an
- ebuild path. #84102
-
-*portage-2.0.51.19 (04 Mar 2005): Maintainence Release
-
- 02 Feb 2005; Jason Stubbs <jstubbs@gentoo.org> pym/portage.py: Fixing
- ferringb's fix by adding the full scope of normpath.
-
- 02 Mar 2005; Nicholas Jones <carpaski@gentoo.org> pemerge.py: Add it back
- since I seem to have removed it by changing/refreshing dirs at some point.
-
- 01 Mar 2005; Brian Harring <ferringb@gentoo.org> pym/portage.py: bug 83712,
- a lovely normalized path bug in eclass_cache resulting in false positives on
- eclass cache staleness...
-
- 01 Mar 2005; Brian Harring <ferringb@gentoo.org> bin/prepman: bug 83704,
- prepman getting stupid about gzip'ing already bzip2'd man pages.
-
- 01 Mar 2005; Jason Stubbs <jstubbs@gentoo.org> bin/emerge: Expanded
- syslist during the unmerge phase so that all packages listed in system
- are correctly checked and warned against. #83670
-
-*portage-2.0.51.18 (28 Feb 2005): Maintainence Release
-
- 28 Feb 2005; Jason Stubbs <jstubbs@gentoo.org> pym/portage.py: Added a
- -* to updated FEATURES before saving back to the env. #79566
-
- 27 Feb 2005; Jason Stubbs <jstubbs@gentoo.org> pym/portage.py: Added back
- sorting of the final list from portdbapi.cp_all() so that emerge regen
- is ordered correctly.
-
-*portage-2.0.51.17 (26 Feb 2005): Maintainence Release
-
- 26 Feb 2005; Nicholas Jones <carpaski@gentoo.org> *: Darwin userland patch
- from Kito (bug 82312). Adjusted a couple bits like the file was installed
- with message and etc-update's USERLAND check per Jason suggestion.
-
- 26 Feb 2005; Nicholas Jones <carpaski@gentoo.org> portage_data.py: Added
- the 'gnu' suffix check for the BSD/GNU people. (bug 80018)
-
- 26 Feb 2005; Nicholas Jones <carpaski@gentoo.org> pym/xpak.py: Added in a
- method to get both the data and index segments for use with the other memory
- based functions.
-
- 21 Feb 2005; Brian Harring <ferringb@gentoo.org> pym/portage.py: So yeah.
- Don't fool with the eclass_cache.porttree ordering unless you understand that
- it -must- match bash's inherit order (where overlays override portdir).
-
- 15 Feb 2005; Jason Stubbs <jstubbs@gentoo.org> emerge: Disabled the cache
- update output when using --quiet. #81678
-
- 15 Feb 2005; Jason Stubbs <jstubbs@gentoo.org> make.conf*: Added
- warning with regard to changing CHOST. #81007
-
- 13 Feb 2005; Jason Stubbs <jstubbs@gentoo.org> make.conf*: Added warning
- with regard to changing PORTDIR and /etc/make.profile.
-
- 13 Feb 2005; Jason Stubbs <jstubbs@gentoo.org> portage.py: Fixed the global
- features usage in the config class and made sure that any adjustments are
- propogated back to the env.
-
- 13 Feb 2005; Jason Stubbs <jstubbs@gentoo.org> chflags.c: Replaced stat
- call with an lstat call in lhasproblems as the calling code expects the
- function to operate directly on symlinks.
-
-*portage-2.0.51.16 (06 Feb 2005): Maintainence Release + BSD chflags
-
- 06 Feb 2005; Nicholas Jones <carpaski@gentoo.org> portage.py, src/chflags:
- Addition of BSD chflags support (from spb@gentoo.org). Wrapped all the
- calls within a check for the modules existance so we can reduce any problems
- due to a failing or missing module. Internally renamed it bsd_chflags.
-
- 06 Feb 2005; Jason Stubbs <jstubbs@gentoo.org> bin/ebuild.sh: Completed
- renaming of "maketest" to "test". RESTRICT supports both. #77446
-
- 06 Feb 2005; Jason Stubbs <jstubbs@gentoo.org> pym/output.py: Added checking
- of whether stderr is a tty before changing xterm titles. #73824
-
- 05 Feb 2005; Jason Stubbs <jstubbs@gentoo.org> bin/emerge: Added removal of
- --tree and --changelog upon a restart of emerge. Added --ask support to
- emerge metadata.
-
- 05 Feb 2005; Jason Stubbs <jstubbs@gentoo.org> pym/portage.py: Removed the
- requisite that a profile must depend on ">=sys-apps/portage-2.0.51" to be
- able to use package.mask.
-
- 05 Feb 2005; Jason Stubbs <jstubbs@gentoo.org> bin/repoman pym/portage.py:
- Removed the disabling of PORTDIR_OVERLAY from portage.py and added code to
- repoman to set PORTDIR to the current directory's repo unless it doesn't
- contain a profiles directory. If that is the case, PORTDIR is kept and
- PORTDIR_OVERLAY is set to the current overlay -only-.
-
- 03 Feb 2005; Brian Harring <ferringb@gentoo.org> pym/portage.py: bug 80506,
- fixed RESTRICT="(no|)userpriv".
-
- 31 Jan 2005; Brian Harring <ferringb@gentoo.org> pym/portage.py: Spanky pointed
- a bug assuming the intermediate directories for WORLD_FILE existed already.
-
- 30 Jan 2005; Brian Harring <ferringb@gentoo.org> bin/ebuild.sh: Update from
- eradicator, bug #75420. Multilib stuff. :)
-
- 30 Jan 2005; Brian Harring <ferringb@gentoo.org> bin/prepman: Fixing bug
- #79788. Some screwy syntax for a find call was screwing up gzip'ing of man
- pages for osx boxes.
-
- 30 Jan 2005; Jason Stubbs <jstubbs@gentoo.org> emerge emerge.1: Added short
- -N option for --newuse. Moved virtuals updating for packages that are about
- to be installed so that the update happens after the virtuals within that
- package's direct dependencies are resolved. Made the blocking message
- slightly easier to understand. Added a suggestion to run emerge with --newuse
- before running depclean.
-
- 30 Jan 2005; Jason Stubbs <jstubbs@gentoo.org> etc-update: Removed the
- unsetting of various options that happens when using graphical mode.
-
- 30 Jan 2005; Jason Stubbs <jstubbs@gentoo.org> quickpkg: Added -h and --help
- options.
-
- 30 Jan 2005; Jason Stubbs <jstubbs@gentoo.org> repoman: Added patch to call
- xmllint directly on metadata.xml rather than piping it. This prevents the
- need (and possible failure) to escape quote characters.
-
- 30 Jan 2005; Jason Stubbs <jstubbs@gentoo.org> portage.py: Fixed a bug with
- calculations using PRELINK_PATH_MASK where paths matching those in
- PRELINK_PATH_MASK were not removed.
-
- 28 Jan 2005; Brian Harring <ferringb@gentoo.org> portage_db_flat.py: removed
- the portage_locks locking that was added, using a different strategy reliant
- on rename. Basically, for a pull- you open the file. If that fails, then
- the cache lacks that entry. If it succeeds, you do your stat calls (to get
- mtime) against the -file handle-, via fstat. You can read from that inode
- without issue, but do not assume that that inode is the still accessible via
- another open call (it may've been updated since). Pushes are accomplished
- via writing everything to a temp file, then renaming that file to the correct
- name- chmod/chown/utime calls need to be done prior to the rename.
- So, if a pull is ongoing while a push starts up and finishes, the pull still
- is accessing the old cache entry- worst case scenario, it goes and reupdates
- the cache entry. This is acceptable, since the overhead from using a seperate
- lockfile is much more costly (in some cases, a difference of 3m to 20m).
- Normal cases, roughly 13% improvement.
-
- 24 Jan 2005; Nicholas Jones <carpaski@gentoo.org> portage.py: Fixed the
- fetch/nofetch RESTRICT bug introduced in -r14. Completely disable overlay
- when calling from repoman.
-
- 24 Jan 2005; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Test print out
- fix.
-
- 19 Jan 2005; Brian Harring <ferringb@gentoo.org> ebuild.sh: Quick change to
- features=autoconfig, avoiding an extra subshell.
-
-*portage-2.0.51-r14 (15 Jan 2005): Stable Candidate
-
- 18 Jan 2005; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Unset IFS
- after entering the for loop for bashrc's. The IFS setting applies to the
- source command and messes up the assumptions about bash scripts and IFS.
-
- 18 Jan 2005; Jason Stubbs <jstubbs@gentoo.org> portage.py emerge-webrsync:
- Fixed the bash syntax error in emerge-webrsync as per #77941. Fixed the
- blocking virtuals bug as per #78201. Removed the warning about negations
- and reapplied negations to profile virtuals.
-
- 17 Jan 2005; Nicholas Jones <carpaski@gentoo.org> portage.py: Prevented
- negations from applying in virtuals and provided a warning.
-
- 16 Jan 2005; Nicholas Jones <carpaski@gentoo.org> repoman: Fix to match up
- to the namespace changes in portage.
-
- 16 Jan 2005; Nicholas Jones <carpaski@gentoo.org> portage.py: Patch from a
- bug report on a traceback that occurs when re-digesting a file without any
- distfiles but with a preexisting digest.
-
- 16 Jan 2005; Nicholas Jones <carpaski@gentoo.org> *.py: Added a cvs_id_str
- to each portage module so we can account for them easily. Added printing
- of those versions on a 'emerge -d info'. Cleaned up the * imports in the
- modules so there shouldn't be any namespace pollution left.
-
-*portage-2.0.51-r13 (15 Jan 2005): Testing for virtuals changes
-
- 15 Jan 2005; Nicholas Jones <carpaski@gentoo.org> portage.py: Adjusted the
- virtuals handling into an advanced for of what Jason started using promotion
- instead of blanket settings. Removed the user_profile dir completely when
- using repoman via a new user_profile_dir variable that is stored on creation.
- Split the getvirtuals() call into two pieces so that setinst() can also
- recreate the virtuals properly when modifying the vartree.
-
- 15 Jan 2005; Nicholas Jones <carpaski@gentoo.org> portage_util.py: Modified
- the stack_dictlist function to handle '-*' and do it in the proper order.
-
- 13 Jan 2005; Nicholas Jones <carpaski@gentoo.org> portage.py, emerge:
- Adjusted the version handling for portage so that we can use CVS versions
- and ignore the restarts when using livecvsportage.
-
- 13 Jan 2005; Nicholas Jones <carpaski@gentoo.org> repoman: Added the
- portage/repoman version used to the commit messages.
-
- 13 Jan 2005; Jason Stubbs <jstubbs@gentoo.org> pym/portage.py: Fixed the
- reverse ordering of profile virtuals before installed virtuals stacking.
- Turn a reference assignment into a copy to fix a bug where digesting would
- not include files that were already fetched previously. Added information
- about which package is causing an error to some aux_get() errors.
-
-*portage-2.0.51-r11 (13 Jan 2005): Cleanup
-
- 13 Jan 2005; Nicholas Jones <carpaski@gentoo.org> cnf/*: Updates for the
- OSUOSL -> distfiles.gentoo.org changeover. Addition of PORTAGE_TMPFS to
- the config files.
-
- 13 Jan 2005; Nicholas Jones <carpaski@gentoo.org> portage.py: Fix for a
- typo in the primaryuri index.
-
-*portage-2.0.51-r10 (11 Jan 2005): Cleanup
-
- 12 Jan 2005; Nicholas Jones <carpaski@gentoo.org> repoman: Fixed the
- namespace collision in the time module and calls.
-
- 11 Jan 2005; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Fixes for the
- libdir patch -- libtool can't handle "usr//lib".
-
- 10 Jan 2005; Nicholas Jones <carpaski@gentoo.org> portage.py: Made maketest
- enable 'test' in features. Marked a couple more negative flags for restrict.
- Made tmpfs checking a little smarter and let us enable /dev/shm by default.
-
- 10 Jan 2005; Nicholas Jones <carpaski@gentoo.org> output.py: Fixed the
- colors and overscore problems. Added a larger section and better code
- for generating colors and effects -- Unused at the moment. Added a patch
- that included screen title handling.
-
- 10 Jan 2005; Nicholas Jones <carpaski@gentoo.org> repoman: Rewrite the
- datetime code to use time.gmtime() instead so python2.2 works still.
-
- 10 Jan 2005; Nicholas Jones <carpaski@gentoo.org> regenworld: Made it work
- with cascade profiles by using the proper system reference.
-
- 10 Jan 2005; Nicholas Jones <carpaski@gentoo.org> emerge: Fix for the gcc
- hardened version printing. Moved the stdin close on rsync down below the
- rsync part to allow --ask to work.
-
- 10 Jan 2005; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Die if unpack
- is not given any files to unpack. Modification for the no_inst stuff to be
- less crazy with the echo redirection stuff.
-
- 08 Dec 2005; Jason Stubbs <jstubbs@gentoo.org> ebuild.sh: Moved the "true" to
- the end of the dyn_clean() function.
-
- 06 Dec 2005; Jason Stubbs <jstubbs@gentoo.org> ebuild.sh: Added eradicator's
- multilib-strict patch, which allows checking and failing if binaries are
- about to be installed into an incorrect directory. The feature is configured
- via the MUTLILIB_STRICT_DIRS and MULTILIB_STRICT_DENY and has no effect if
- either they are not defined or the feature is not enabled.
-
- 04 Dec 2005; Jason Stubbs <jstubbs@gentoo.org> ebuild.sh: Added removal of
- the .packaged and .tested files to the clean stage. Adjusted cleaning of the
- temp dir to include the directory as well, rather than only the files
- therein. Added a check and appropriate removal of the entire scratch dir if
- it is empty at the end of the clean phase.
-
- 04 Dec 2005; Jason Stubbs <jstubbs@gentoo.org> portage.py: Turned the
- primaryuri global counter into separate counters per file, so that urls
- are inserted at the appropriate places.
-
- 04 Dec 2005; Jason Stubbs <jstubbs@gentoo.org> portage.py: Adjusted virtuals
- stacking so that /etc/portage/virtuals is stacked on top of the profile
- virtuals (including the custom /etc/portage/profile), reordered the list
- of installed virtuals to match that of the combined profile virtuals and
- then stacked the installed virtuals on top of those. Also added back the
- deprecation notice.
-
- 04 Dec 2005; Jason Stubbs <jstubbs@gentoo.org> emerge: Added code to
- "emerge info" to summarise variables that aren't set rather than hide them.
-
- 02 Dec 2005; Jason Stubbs <jstubbs@gentoo.org> dolib ebuild.sh preplib:
- Applied eradicator's libdir patch for multilib support. #75420
-
- 02 Dec 2005; Jason Stubbs <jstubbs@gentoo.org> ebuild.sh: Added support for
- INSTALL_MASK. #67190
-
- 02 Dec 2005; Jason Stubbs <jstubbs@gentoo.org> portage.py portage_util.py:
- Added support for the "source" keyword to make.conf. #74842
-
- 02 Dec 2005; Jason Stubbs <jstubbs@gentoo.org> ebuild.sh: Adjusted autoconfig
- to work on ${WORKDIR} instead of ${S} and modified output. #72360
- Added notice on econf failure to include config.log (if it exists) into any
- bug report. #75268 Don't prepend "/" to CONF_LIBDIR if it already begins
- with "/" #75523
-
- 02 Dec 2005; Jason Stubbs <jstubbs@gentoo.org> emerge: Don't update info dirs
- if noinfo FEATURE is enabled. #76378
-
- 02 Dec 2005; Jason Stubbs <jstubbs@gentoo.org> emerge: Set the processing of
- profiles/info_vars to only include those which have been set to some value.
- #75920
-
- 02 Dec 2005; Jason Stubbs <jstubbs@gentoo.org> repoman: Modified header check
- to use the year of the mtime of the ebuild rather than the current year.
-
- 24 Dec 2004; Jason Stubbs <jstubbs@gentoo.org> portage.py: Removed assumption
- that /etc/make.profile/make.defaults must exist for configdict["defaults"] to
- be populated (and thus ARCH prepended to USE).
-
- 23 Dec 2004; Jason Stubbs <jstubbs@gentoo.org> portage.py: Adjusted fetching
- on digest creation such that any preexisting digest entries are used, but
- fetching and digesting occurs for all SRC_URI entries otherwise.
-
- 22 Dec 2004; Jason Stubbs <jstubbs@gentoo.org> portage.py: Removed the
- virtuals deprecation notice as the "preferred" location does not yet
- provide the required behaviour.
-
- 14 Dec 2004; Nicholas Jones <carpaski@gentoo.org> doins: Fix and notices
- for $D handling.
-
- 14 Dec 2004; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: spb's chflags
- adjustment for dyn_clean. stat_perms() overhaul for bsd and ppc.
-
- 14 Dec 2004; Nicholas Jones <carpaski@gentoo.org> webrsync: TMPDIR is
- used properly and a tar-failure fix.
-
- 14 Dec 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: RESTRICT
- changes -- nomirror and mirror are now the same (use 'mirror') -- lmirror
- is a temporary way to sidestep mirror restrictions -- primaryuri makes
- portage attempt to retrieve from the source mirror before falling back to
- other mirrors (corrects behavior some impose for 'mirror').
-
- 14 Dec 2004; Nicholas Jones <carpaski@gentoo.org> portage_db_flat.py: Do
- not fail when we're on a readonly FS and we're only reading (lockfiles).
-
- 09 Dec 2004; Marius Mauch <genone@gentoo.org> bin/ebuild.sh:
- change LDFLAGS to CFLAGS for the setXid check.
-
- 05 Dec 2004; Nicholas Jones <carpaski@gentoo.org> pym/portage.py: swegener
- made quite a find in portdbapi.cpv_exists, the comparison on the tuple
- returned by findname2 always evaluated true.
-
-*portage-2.0.51-r8 (03 Dec 2004): Cleanup
-
- 03 Dec 2004; Nicholas Jones <carpaski@gentoo.org> repoman: A patch to
- handle stray digests.
-
- 03 Dec 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Patches to
- fix the matching bug #73136 and the '-atom' profile bug #73167.
-
-*portage-2.0.51-r7 (01 Dec 2004): Cleanup
-
- 01 Dec 2004; Nicholas Jones <carpaski@gentoo.org> sandbox: updated with
- ferringb's code for the pids file overrun issue.
-
- 01 Dec 2004; Jason Stubbs <jstubbs@gentoo.org> repoman: Reverted repoman
- profile selection to previous behaviour. Kept verbosity, but modified
- messages according to the behaviour.
-
-*portage-2.0.51-r5/6 (30 Nov 2004): Cleanup
-
- 29 Nov 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Added a
- (backport?) missing return to treewalk.
-
- 29 Nov 2004; Nicholas Jones <carpaski@gentoo.org> cnf/*: Removed the
- prozilla references. Adjusted the default mirrors to a gentoo round-
- robin.
-
- 29 Nov 2004; Nicholas Jones <carpaski@gentoo.org> portage_exec.py: Modified
- the magic constants to be os.access calls.
-
- 29 Nov 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: A few XXX
- comments. Removed 'aumtime' from the global space. Removed the virtuals
- code that created an empty and unused virtuals file in /var/cache/edb.
-
- 29 Nov 2004; Nicholas Jones <carpaski@gentoo.org> emerge: Added a few XXX
- comments to bad code/calls. Adjusted emerge info so that it can read more
- variables and atoms to print out from the portage tree.
-
- 29 Nov 2004; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Fixed/Backport
- the changes to stat_perms so that it works. Fixed the read commands.
-
- 10 Nov 2004; Brian Harring <ferringb@gentoo.org> bin/ebuild.sh:
- dyn_install was attempting to be helpful and chown files owned
- by portage, resulting in suid/sgid being stripped. bug #56129.
-
- 07 Nov 2004; Brian Harring <ferringb@gentoo.org> portage.py: buggy
- logic in listdir w/ ignorecvs, rewrote. (#70170).
-
- 07 Nov 2004; Jason Stubbs <jstubbs@gentoo.org> dispatch-conf: Moved back
- to the version of dispatch-conf that was released (due to python-2.2
- compatibility) and fixed a small type in that version. #70282
-
-*portage-2.0.51-r4 (05 Nov 2004): Internal Release
-
- 05 Nov 2004; Jason Stubbs <jstubbs@gentoo.org> doins: Added vapier's fix
- for doins changing existing directory permissions. #69896
-
- 04 Nov 2004; Jason Stubbs <jstubbs@gentoo.org> tbz2tool.c: Added vapier's
- fix for tbz2tool on ia64. #70001
-
- 03 Nov 2004; Brian Harring <ferringb@gentoo.org> portage_dep.py: Bug in
- paren_reduce where it was returning a tuple (supposed to return a list).
-
- 03 Nov 2004; Jason Stubbs <jstubbs@gentoo.org> dispatch-conf: Simplified
- the temp file creation and usage.
-
- 03 Nov 2004; Jason Stubbs <jstubbs@gentoo.org> portage.py: Fixed issues
- with PROVIDE calculation by flattening the dep array. #32114
-
- 02 Nov 2004; Jason Stubbs <jstubbs@gentoo.org> portage.py: Updated
- ExtractKernelVersion() to support 2.6.9's new localversion. #67804
- Reordered a cache update and a return statement so that the cache is
- actually updated in ververify. #69523
-
- 02 Nov 2004; Jason Stubbs <jstubbs@gentoo.org> ebuild.sh: Turned off
- shell option extglob as it interferes with the g++ QA interceptor. #69690
-
- 02 Nov 2004; Jason Stubbs <jstubbs@gentoo.org> portage.py: Fixed a bug
- where a broken symlink blocking the installation of a regular file would
- cause a traceback. #69672
-
- 02 Nov 2004; Jason Stubbs <jstubbs@gentoo.org> quickpkg: Applied vapier's
- fix for processing file names that contain spaces. #36997 Applied
- vapier's fix for parallel quickpkg's writing to a single log file. #37270
-
- 29 Oct 2004; Jason Stubbs <jstubbs@gentoo.org> emerge: Fixed issue where
- emerge adds package to world when --onlydeps is specified. #69260
- Fixed issue where emerge was adding packages to world when --uprade is
- specified even if the package is already installed. #69287 Fixed logic
- in path/to/ebuild check so that symlinks do not cause false positives.
-
- 29 Oct 2004; Jason Stubbs <jstubbs@gentoo.org> dispatch-conf: Added usata's
- fix for mac-os compatibility. #69304
-
- 29 Oct 2004; Jason Stubbs <jstubbs@gentoo.org> getbinpkg.py: Reversed the
- logic of ftp passive-mode selection. #69371
-
- 29 Oct 2004; Brian Harring <ferringb@gentoo.org> portage.py: config.reset
- was pruning vars out of backupenv on each reset call. #69388
-
- 28 Oct 2004; Jason Stubbs <jstubbs@gentoo.org> emergehelp.py: Fixed up
- some incorrect descriptions. Added documentation for new and changed
- functionality. Removed documentation for deprecated options.
-
- 28 Oct 2004; Jason Stubbs <jstubbs@gentoo.org> getbinpkg.py: Added missing
- substitution of ${FILE} in resume command. #69068
-
- 28 Oct 2004; Jason Stubbs <jstubbs@gentoo.org> repoman: Fixed aux_get error
- due to calling getfetchlist() when an ebuild no longer exists. #69051
-
- 28 Oct 2004; Jason Stubbs <jstubbs@gentoo.org> emerge: Removed a duplicate
- loop from the userquery function. #69020
-
- 27 Oct 2004; Jason Stubbs <jstubbs@gentoo.org> dispatch-conf: Moved temp file
- creation into a safe directory under /tmp. Made the log file a configurable
- option that is disabled by default.
-
- 26 Oct 2004; Jason Stubbs <jstubbs@gentoo.org> portage.py: Added a check to
- portdbapi.fetchcheck to account for a missing digest.
-
- 26 Oct 2004; Brian Harring <ferringb@gentoo.org> portage_exec.py, bin/ebuild:
- bug with the path lookup code (added find_binary func for upcoming additions),
- and bin/ebuild was bailing due to a missing portage_util import.
-
- 25 Oct 2004; Jason Stubbs <jstubbs@gentoo.org> portage.py: Wrapped entire
- lock-holding section of fetch() in a try-finally to ensure that the lock
- file gets released regardless of exception.
-
- 25 Oct 2004; Jason Stubbs <jstubbs@gentoo.org> etc-update: Added patch
- from #48218 to continue on to next file when answering "no" to the
- interactive replace prompt of menu option -3.
-
- 25 Oct 2004; Jason Stubbs <jstubbs@gentoo.org> emerge: Fixed the space/tab
- usage through the spinner_msgs definition. Removed COMPILER from emerge info
- output.
-
- 25 Oct 2004; Jason Stubbs <jstubbs@gentoo.org> portage.py: Converted
- config.pkeywordsdict from {atom:[keyword]} to {cp:{atom:[keyword]}} to
- prevent a lot of unnecessary calculation.
-
- 24 Oct 2004; Jason Stubbs <jstubbs@gentoo.org> emerge: Added a percentage
- counter to the cache update phase based on patch from #68694
-
- 24 Oct 2004; Jason Stubbs <jstubbs@gentoo.org> portage.py:
- s/macos/ppc-macos/ change.
-
- 24 Oct 2004; Jason Stubbs <jstubbs@gentoo.org> portage_locks.py
- portage_db_flat.py: Fixed a couple of race conditions with regard to
- stating files.
-
- 23 Oct 2004; Jason Stubbs <jstubbs@gentoo.org> portage.py portage_util.py:
- Added syntax checking for package.mask, package.unmask and packages files.
-
- 23 Oct 2004; Jason Stubbs <jstubbs@gentoo.org> dohtml: Reimplemented the
- option processing to remove the dependency on optparse.
-
- 23 Oct 2004; Jason Stubbs <jstubbs@gentoo.org> repoman: Made the invalid
- profile slightly more intelligable and got rid of the duplicate. Moved
- the profile KEYWORDS.invalid failure so that it only occurs if the
- corresponding KEYWORD is valid.
-
- 23 Oct 2004; Jason Stubbs <jstubbs@gentoo.org> portage.py: Added settings
- parameter to dep_virtual as the virtuals can now change during dep graph
- creation. #68220
-
- 23 Oct 2004; Jason Stubbs <jstubbs@gentoo.org> repoman: Changed repoman to
- use the first stable profile found for each arch listed in profiles.desc
- rather than the last profile listed. Added notification on any invalid
- profile in profiles.desc. Made exception into a repoman error where a
- ebuild is using a keyword for which no profile can be found.
-
- 23 Oct 2004; Jason Stubbs <jstubbs@gentoo.org> repoman: Fixed traceback in
- repoman on invalid LICENSE syntax.
-
- 22 Oct 2004; Brian Harring <ferringb@gentoo.org> ebuild.sh: silencing use/has.
-
- 22 Oct 2004; Jason Stubbs <jstubbs@gentoo.org> emerge: Added python version
- to "emerge info" output.
-
-*portage-2.0.51-r2 (20 Oct 2004): Everyone loves finding bugs in stable!
-
- 20 Oct 2004; Nicholas Jones <carpaski@gentoo.org> emerge: Fixed the result
- checking when ebuilds are called -- this removes the continuing-after failure
- big discovered by AMD64 today.
-
- 20 Oct 2004; Jason Stubbs <jstubbs@gentoo.org> emerge: Fixed a bug where
- users were warned against emerging by path when a file/dir exists of the
- same name of the package to be merged. #68372 Added an warning with delay
- when emerging an ebuild that is masked. Added a check on PORTAGE_GPG_DIR
- which removes "gpg" from FEATURES on an invalid settings. #68387
-
-*portage-2.0.51 (20 Oct 2004): Everyone loves stable!
-
- 19 Oct 2004; Nicholas Jones <carpaski@gentoo.org> portage.5: patch included
- to fix a few typos.
-
- 19 Oct 2004; Nicholas Jones <carpaski@gentoo.org> g-cpan.pl patch included
- that doesn't recreate inherently-created ebuilds.
-
- 20 Oct 2004; Jason Stubbs <jstubbs@gentoo.org> repoman: Added check for
- digest entries that aren't used within the corresponding ebuild's SRC_URI.
-
- 20 Oct 2004; Jason Stubbs <jstubbs@gentoo.org> emerge: Added support for
- EMERGE_WARNING_DELAY defaulting it to 10. Changed all the hardcoded delays
- to use it. Needed for the catalyst guys as it includes a number of unmerges
- of system packages.
-
- 20 Oct 2004; Jason Stubbs <jstubbs@gentoo.org> emerge: Changed the
- /path/to/ebuild warning. Added a check on what ebuild is chosen and
- a failure if it doesn't match what was specified. Added the same for
- /path/to/tbz2.
-
-*portage-2.0.51_rc10 (19 Oct 2004): Potential Final #1.
-
- 19 Oct 2004; Nicholas Jones <carpaski@gentoo.org> emerge: Visual acuity
- enhancements. Fixed all the spaces in the option handling.
-
- 19 Oct 2004; Nicholas Jones <carpaski@gentoo.org> emerge.1: Updated for
- 'f' and 'S'.
-
- 19 Oct 2004; Nicholas Jones <carpaski@gentoo.org> getbinpkg.py: Potential
- fix for a missing dict key.
-
- 19 Oct 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Prelink
- tempfile per pid.
-
- 18 Oct 2004; Nicholas Jones <carpaski@gentoo.org> *.py: Added a protocol
- change for cPickles to make them more efficient (binary).
-
- 18 Oct 2004; Nicholas Jones <carpaski@gentoo.org> sandbox: Fixes for some
- incorrect indexes that cause some boxes to have very random issues.
-
- 18 Oct 2004; Nicholas Jones <carpaski@gentoo.org> portage_locks.py: Reverted
- the lock code to the lockf()+hardlock version as the new code has some
- inexplicable incompatibility with NFS.
-
- 18 Oct 2004; Nicholas Jones <carpaski@gentoo.org> portage_exception.py: Added
- a couple exceptions pretaining specifically to packages and digests.
-
- 18 Oct 2004; Nicholas Jones <carpaski@gentoo.org> portage_checksum.py: Now
- takes care of missing files better for the partial-digest handling.
-
- 18 Oct 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Fixed up some
- of the output lines to use writemsg instead of print. Adjusted the fetch
- code to handle partial-digests. digestCreate can substitute in old digest
- entries now for missing files -- the assumption is that they are correct.
- Portage no longer downloads all parts when FEATURES=cvs is enabled, but
- does require a complete digest to commit, whether it was created piecewise
- or en masse. Added 'fetch_check' to portdbapi which helps with with emerge's
- pretend output for Fetch-Restriction.
-
- 18 Oct 2004; Nicholas Jones <carpaski@gentoo.org> repoman: Added in support
- for partial-digests -- It complains, downloads, and adjusts. Sorted the QA
- categories to make them easier to scan.
-
- 18 Oct 2004; Nicholas Jones <carpaski@gentoo.org> emerge: Added -F to force
- all URIs to be downloaded. Added in a green, lowercase 'f' when the fetch
- restriction is satisfied for pretend. Added Kevin Quinn's patch for some
- prepstrip sanity and friendliness regarding TEXTRELs.
-
- 18 Oct 2004; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Added a message
- about use/useq/usev. CCACHE_DIR and CCACHE_SIZE are now properly exported.
-
- 17 Oct 2004; <genone@gentoo.org> pym/portage.py:
- Added a followSymlinks parameter to listdir() as otherwise it gets stuck in
- an infinite loop when it encounters self-referencing symlinks, this behavior
- was exhibited by the collision-protect feature.
-
- 17 Oct 2004; Brian Harring <ferringb@gentoo.org> portage-locks.py: Tweaked
- unlockfile, no point in attempting to relock the file for deletion, if you're
- not deleting the lockfile (basically moved locking w/in if unlinkfile).
-
- 17 Oct 2004; Jason Stubbs <jstubbs@gentoo.org> emerge portage_locks.py:
- Added more information to the --inject deprecation notice. Removed the sleep
- call from unlockfile as it severely affects performance (apparently only
- under some schedulers)
-
- 16 Oct 2004; Brian Harring <ferringb@gentoo.org> portage_locks.py: Corrected
- a bug involving unlocking + lockf.
-
- 12 Oct 2004; Brian Harring <ferringb@gentoo.org> portage_exec.py: spawn_sandbox
- wasn't passing the opt_name down to spawn, it now does.
-
- 11 Oct 2004; Jason Stubbs; <jstubbs@gentoo.org> emerge: Reverted back to
- previous command line parsing code as --search options were being mishandled.
-
- 11 Oct 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Fix for the
- SHA1 digest slipout and the exec missing/broken binary traceback.
-
- 11 Oct 2004; Nicholas Jones <carpaski@gentoo.org> pym/*: See below.
-
- 11 Oct 2004; Jason Stubbs <jstubbs@gentoo.org> bin/*: Added catching and
- propogating of SystemExit exception to all blanket exception handlers.
-
- 11 Oct 2004; Jason Stubbs <jstubbs@gentoo.org> repoman: Fixed repoman LICENSE
- check to accept || () conditions.
-
-*portage-2.0.51_rc8 (10 Oct 2004): RC + Lock cleanup, Happy RC #4
-
- 10 Oct 2004; Nicholas Jones <carpaski@gentoo.org> emerge: Added in close()
- calls to ensure that control-C gets intercepted by portage instead of the
- children. Added in a re-raise for the SysExit exception inside of the
- regen and metadata targets.
-
- 10 Oct 2004; Nicholas Jones <carpaski@gentoo.org> cnf/make.conf*:
- benno@nietvergeten.nl's touchups to make.conf files.
-
- 10 Oct 2004; Nicholas Jones <carpaski@gentoo.org> sandbox/*: Includes one
- of solar's patches to fix up potential holes in sandbox.
-
- 10 Oct 2004; Nicholas Jones <carpaski@gentoo.org> portage_exec.py: Added
- a cleanup routine.
-
- 10 Oct 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: A could cwd
- fixes and removed the compat-writing for digests.
-
- 10 Oct 2004; Nicholas Jones <carpaski@gentoo.org> md5check/mirror: brought
- them more up to date for the digest changes.
-
- 10 Oct 2004; Nicholas Jones <carpaski@gentoo.org> emerge: Moved the imports
- around to ensure that portage.py gives output on failure instead of emerge
- just dying. Added '-1' as a short for oneshot. Added shorter messages for
- the titlebar. Fixed signal handling more -- Emerge sets up a handler that
- calls to portageexit() before quitting normally.
-
- 10 Oct 2004; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Changed around
- the tarvars to make them STAR friendly. Added ECONF_SOURCE as a way to move
- econf out of the source directory and into a new build directory -- it
- defines the path to configure, not the command. Added CTARGET support. Added
- an rm for the infodir entries installed by autoconf so they don't kill what
- portage regens and vice versa.
-
- 10 Oct 2004; Nicholas Jones <carpaski@gentoo.org> doins: Added Spanky's
- recursion patch.
-
- 10 Oct 2004; Nicholas Jones <carpaski@gentoo.org> dolib: Added Spanky's
- symlink fix and condensed dolib*.
-
- 10 Oct 2004; Nicholas Jones <carpaski@gentoo.org> doman: Added Spanky's
- i18n patch.
-
- 08 Oct 2004; Brian Harring <ferringb@gentoo.org> portage_exec.py: Removed
- the portage_exec.spawn_bash call for tee logging- instead, transferred in
- some code that does path lookups (closer to the older portage.spawn call).
- Path lookup by default is on, but can be disabled via path_lookup=False.
-
- 08 Oct 2004; Brian Harring <ferringb@gentoo.org> portage_locks.py: Reverted
- to using flock by default- if it fails (unavailable), -then- use lockf, then
- hardlink.
-
- 07 Oct 2004; Jason Stubbs <jstubbs@gentoo.org> portage.py: Updated
- portdbapi.getfetchsizes function for new digestParseFile return values.
-
- 05 Oct 2004; Jason Stubbs <jstubbs@gentoo.org> quickpkg: Fixed bug whereby
- creating a package from within /var/db/pkg/cat and specifying pkg would
- create a broken package.
-
- 05 Oct 2004; Jason Stubbs <jstubbs@gentoo.org> emerge: Modified output for
- slotted installations. #26139
-
- 05 Oct 2004; Jason Stubbs <jstubbs@gentoo.org> emerge: Refactored argv
- processing a little bit and made "emerge rsync" to emerge rsync with notice
- and "emerge --rsync" to emerge --sync with notice.
-
- 05 Oct 2004; Nicholas Jones <carpaski@gentoo.org> portage_util.py: grabfile
- now handles a compat_level option for comment-based compatability changes --
- This feature is for migration only and is thus transitory.
-
- 05 Oct 2004; Nicholas Jones <carpaski@gentoo.org> portage_exception: Added
- a DigestException which is a SignatureException.
-
- 05 Oct 2004; Nicholas Jones <carpaski@gentoo.org> perform_checksum.py: Added
- a perform_all() and verify_all() functions that handle the new dict of hashes
- that digestParse returns -- It creates hashes for all the listed/known
- formats or verifies them all returning a tuple of ok,reason.
-
- 05 Oct 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Cleared up
- a few system-package import-alls (stat,commands) -- If it breaks something,
- they can fix their code -- They are standard modules. Removed the import
- for select, as it appears nowhere in usage. Changed the portage_data import.
- Added exithandler() back into usage, it was disabled -- Also corrected it's
- handling. Unified the digest-parsing code and made it into an intelligable
- dict instead of the fixed format. Merged digest functions and added SHA1
- (arbitrary) handling in a new COMPAT mode using comments until we get the
- handling transitioned into common usage. digestCreateLines() handles the
- compatibility line values, and grabfile() has a compat-level handler.
-
- 05 Oct 2004; Nicholas Jones <carpaski@gentoo.org> man/*: Random touchups.
-
- 04 Oct 2004; <jstubbs@gentoo.org> pym/portage.py: Added fix for config
- protection failure when destination is a symlink. #13007
-
- 04 Oct 2004; <jstubbs@gentoo.org> bin/repoman: Added detection of multiple
- overlays to repoman.
-
- 04 Oct 2004; <jstubbs@gentoo.org> bin/repoman: Added repoman check
- for DEPEND-syntax following LICENSEs.
-
- 03 Oct 2004; <genone@gentoo.org> pym/emergehelp.py:
- Add --metadata documentation to --help output.
-
- 03 Oct 2004; <genone@gentoo.org> cnf/*, man/emerge.1, man/make.conf.5,
- man/portage.5, pym/emergehelp.py:
- Changed documentation to use --action instead of action (bug #2365).
- Also changed ufed references in make.conf to use the correct category.
-
-*portage-2.0.51_rc7 (30 Sep 2004): RC + Lock cleanup, Happy RC #3
-
- 30 Sep 2004; Nicholas Jones <carpaski@gentoo.org> emerge: Rsync fix part 2
- for the distfiles, local, and packages directory unlinks.
-
- 30 Sep 2004; Nicholas Jones <carpaski@gentoo.org> emerge-webrsync: Fix for
- the missing md5sum causing failure to download. Fixed up a couple messages.
- Modified the local rsync line.
-
- 30 Sep 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Make more of
- the chown calls friendly. More output on strange exceptions in aux_get.
-
- 30 Sep 2004; Nicholas Jones <carpaski@gentoo.org> portage_db_template.py:
- Added some more putput to the corruption message.
-
- 29 Sep 2004; Jason Stubbs <jstubbs@gentoo.org> output.py: Added unicode-rxvt
- to the list of legal term types. #65762
-
- 28 Sep 2004; Jason Stubbs <jstubbs@gentoo.org> etc-update: Added patch to
- use gsed on BSD from bug 60721.
-
-*portage-2.0.51_rc6 (26 Sep 2004): RC + Lock cleanup, Happy RC #2
-
- 26 Sep 2004; Nicholas Jones <carpaski@gentoo.org> emerge: Restart fix for
- the -a into execv code.
-
- 26 Sep 2004; Nicholas Jones <carpaski@gentoo.org> portage_locks.py: Fixed
- a traceback for Fat32 users.
-
-*portage-2.0.51_rc5 (26 Sep 2004): RC + Lock cleanup, Happy RC #1
-
- 26 Sep 2004; Nicholas Jones <carpaski@gentoo.org> portage_data.py: Fixed
- the BSD lchown issues.
-
- 26 Sep 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Added the
- selinux secure dirs patch.
-
- 26 Sep 2004; Nicholas Jones <carpaski@gentoo.org> emerge: Modified the
- spinner again, --nospinner provides a basic ticker of one '.' per 100,
- normal spinner is the twirly one, and the FEATURES=candy spinner is a
- scrolly message.
-
- 26 Sep 2004; Nicholas Jones <carpaski@gentoo.org> portage_locks.py: Fixed it.
- Discovered that the hardlocks were failing due to the creation of the lock
- prior to the link operation which was due to the NFS fcntl lock failure.
-
- 26 Sep 2004; Jason Stubbs <jstubbs@gentoo.org> pym/portage.py: Added
- support for per profile package.mask. Included check of packages file
- to ensure that the profile depends on an adequate portage version.
-
-*portage-2.0.51_rc4 (26 Sep 2004): RC + Lock cleanup
-
- 26 Sep 2004; Nicholas Jones <carpaski@gentoo.org> bin/clean_locks: A new
- tool to aid in the maintainence of hardlock-based locks. It can clean
- all locks from a directory or just the ones pertinent to the running host.
-
- 26 Sep 2004; Nicholas Jones <carpaski@gentoo.org> emerge: Updated the
- spinner.
-
- 26 Sep 2004; Nicholas Jones <carpaski@gentoo.org> cnf/*: Added distlocks
- as a default feature. Added comments on distlocks and maketest and gpg.
-
- 26 Sep 2004; Nicholas Jones <carpaski@gentoo.org> portage_locks.py: Lots
- of changes. Corrected the lockfile code to be an IOError. Added some
- helper functions to reduce duplication in the hardlink code. Added a
- callback to a cleanup function registered with atexit to ensure we clean
- locks up on normal terminations. Fixed the code to actually work on most
- NFS systems and hopefully have the fallback (INODE test) working on
- very broken systems. Added a cleanup function that is interfaced through
- the clean_locks script and the registered atexit call.
-
-*portage-2.0.51_rc3 (26 Sep 2004): And we have another Release Candidate!
-
- 26 Sep 2004; Nicholas Jones <carpaski@gentoo.org> emerge: Modified the
- spinner.
-
- 26 Sep 2004; Nicholas Jones <carpaski@gentoo.org> portage_db_flat.py: Fixed
- a lockfile descriptor leak due to duplicated lock calls.
-
- 26 Sep 2004; Nicholas Jones <carpaski@gentoo.org> portage_lock.py: Added
- more cleanup to the lockfile descriptors.
-
-*portage-2.0.51_rc2 (25 Sep 2004): And we have another Release Candidate!
-
- 25 Sep 2004; Nicholas Jones <carpaski@gentoo.org> portage_db_flat.py: Fix
- for typos.
-
- 25 Sep 2004; Nicholas Jones <carpaski@gentoo.org> portage_locks.py: Changed
- all the calls to lockf which wraps fcntl. Made the chown on the locks
- optional -- if it fails, it'll be annoying, but there's a message. Added
- code to perform the hardlink-shuffle which uses hardlinks as a locking
- mechanism (NFSv2 needs this).
-
- 25 Sep 2004; Nicholas Jones <carpaski@gentoo.org> portage_db_flat.py:
- Added locking around the file creation to ensure atomicity.
-
- 25 Sep 2004; Nicholas Jones <carpaski@gentoo.org> portage_localization.py:
- A little spot to provide the '_' function and examples and future code.
-
- 25 Sep 2004; Nicholas Jones <carpaski@gentoo.org> portage_file.py: Added
- the module to contain file/directory functions that are useful. Additional
- function 'makedirs' handles creation of directories with recursive perms.
-
- 25 Sep 2004; Nicholas Jones <carpaski@gentoo.org> getbinpkg.py: Fixed the
- exception handling to not traceback.
-
- 25 Sep 2004; Nicholas Jones <carpaski@gentoo.org> etc-update: Typo fix.
-
- 25 Sep 2004; Nicholas Jones <carpaski@gentoo.org> emerge: Caused the
- binary metadata downloading to be a little more verbose. Fixed the
- sync command's arguments so that it deletes top level files.
-
- 25 Sep 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Added
- Cretin's patch for prelink blacklists. Increased the verbosity of the
- 'mylines' output to let people know what files are affected with nulls.
- Added a workaround for a race condition that somehow exists inside of
- auxget when there is heavy lockfile contention -- Must be a lockfile
- cleanup issue.
-
- 25 Sep 2004; Jason Stubbs <jstubbs@gentoo.org> emerge: Modified env_update
- to always run ldconfig if makelinks is True, in order to ensure that
- missing symlinks are created. Added logic to treewalk to check if package
- is being downgraded and only run env_update with makelinks=False in that
- case. (#54655)
-
- 24 Sep 2004; Jason Stubbs <jstubbs@gentoo.org> emerge: Added the removal
- of --ask from argv when restarting after an emerge of portage. (#47379)
-
- 21 Sep 2004; Jason Stubbs <jstubbs@gentoo.org> portage_util.py: Changed
- varexpand to convert '\'-prepended newline chars to space rather than '\n'.
-
- 21 Sep 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Fixed the
- bintree ebuild locating. Removed the virts_p debug/bug statements.
-
- 21 Sep 2004; Nicholas Jones <carpaski@gentoo.org> portage_checksum: Added
- a fix for checksum tracebacks that tracebacked.
-
- 21 Sep 2004; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Added more
- variables to vardb.
-
- 21 Sep 2004; Jason Stubbs <jstubbs@gentoo.org> emerge: Added (slightly
- modified) patch from bug 64682.
-
- 16 Sep 2004; Jason Stubbs <jstubbs@gentoo.org> portage.py: Modified
- getmaskingstatus() to use settings.prevmaskdict rather than reading packages
- directly in support of cascading profiles.
-
-*portage-2.0.51_rc1 (25 Sep 2004): And we have a Release Candidate!
-
- 15 Sep 2004; Nicholas Jones <carpaski@gentoo.org> ebuild: On merge, disable
- the noauto feature explicitly.
-
- 15 Sep 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Fixed the
- 'missing ebuild' for uninstalls of packages not in the tree.
-
- 15 Sep 2004; Nicholas Jones <carpaski@gentoo.org> man/*: Started adding
- to the documentation in order to break it an make it look funny. Added a
- couple entries for missing concepts like 'inherit' and 'useq' and 'hasq'.
- Cleared up a few things here and there with usage. Added in the metadata
- target for emerge.
-
-*portage-2.0.51_pre24 (14 Sep 2004): Last _pre before docs and rc/stable.
-
- 14 Sep 2004; Nicholas Jones <carpaski@gentoo.org> repoman: Extra handling
- for weird CVS/Repository info on OSX.
-
- 14 Sep 2004; Nicholas Jones <carpaski@gentoo.org> emerge: Fixed a traceback
- in traceback handling where a value was trying to be extracted from an
- exception. Handled the case where a binary package does not have an ebuild
- in a tree or overlay and the verbose overlay output requires it. Change to
- the CVS checkouts for emerge sync, moved the -P immediately after the co.
-
- 14 Sep 2004; Nicholas Jones <carpaski@gentoo.org> portage_exception.py:
- Added in some spacing between related exception groups.
-
- 14 Sep 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Changed a bunch
- of lines unnecessarily to start getting the locale strings ready -- I'll
- probably have bugs due to this because I've been doing dumb things like
- that recently.
-
- 14 Sep 2004; Nicholas Jones <carpaski@gentoo.org> man/*: Fixed ka0ttic's
- email address.
-
- 14 Sep 2004; Brian Harring <ferringb@gentoo.org> bin/ebuild.sh: Removed
- the ${T}/successful logic, it's no longer needed. Existed only for the
- $0 "$@" 2>&1 | tee $PORTAGE_LOG trickery, which is now handled via
- portage_exec.spawn.
-
- 13 Sep 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Fix for empty
- categories/portage-tree causing a traceback.
-
-*portage-2.0.51_pre23 (11 Sep 2004): Fixes and stuff.
-
- 11 Sep 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Another fix
- to the virtuals/use code -- It wasn't using treeVirtuals.
-
-*portage-2.0.51_pre22 (11 Sep 2004): Fixes and stuff.
-
- 11 Sep 2004; Jason Stubbs <jstubbs@gentoo.org> portage.py: Fixed breakage
- in cacheddir changes upon stat'ing a broken symlink.
-
- 11 Sep 2004; Jason Stubbs <jstubbs@gentoo.org> portage.py: Fixed missing
- check in autouse function. Fixed out-of-bounds exception on catpkgsplit
- tuple access in vardbapi.move_ent. Changed to manual stat calls in cacheddir.
-
- 10 Sep 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Removed the
- circular deps of vardbapi and config.
-
-*portage-2.0.51_pre21 (09 Sep 2004): Fixes and stuff.
-
- 09 Sep 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Possible fix
- for autouse() that will prevent recursion and locks.
-
-*portage-2.0.51_pre210 (09 Sep 2004): (pre-pre release for 21)
-
- 09 Sep 2004; Nicholas Jones <carpaski@gentoo.org> archive-conf: Added patch
- so that it runs.
-
- 09 Sep 2004; Nicholas Jones <carpaski@gentoo.org> dispatch-conf: Added patch
- to die when rcs isn't installed but is required by options.
-
- 09 Sep 2004; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Solar's checks
- for suid bind issues.
-
- 09 Sep 2004; Nicholas Jones <carpaski@gentoo.org> fixvardbentries: Updated
- to Jason's current script.
-
- 09 Sep 2004; Nicholas Jones <carpaski@gentoo.org> g-cpan.pl: Added a fix
- to store the ebuilds in a defined overlay.
-
- 09 Sep 2004; Nicholas Jones <carpaski@gentoo.org> output.py: Added all
- xterm* terms to the title-list.
-
- 09 Sep 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Fixed the
- cascade/stack functions -- use.defaults is fully line-incremental now.
- Removed a lot of cruft commented-code. Added an ebuild-mover into the
- entry move functions.
-
- 09 Sep 2004; Nicholas Jones <carpaski@gentoo.org> portage_util.py: Placed
- the lex code into a try/except so we can add the filename into the error
- that is passed back from shlex.
-
- 08 Sep 2004; Brian Harring <ferringb@gentoo.org> portage.py: Modified
- config.__init__(clone=1) so that profiles list is preserved, allowing for
- all profile's bashrc's to be sourced.
-
- 06 Sep 2004; Brian Harring <ferringb@gentoo.org> portage.py: Fixed
- fetch logic for when DISTDIR isn't writable, but the file is fully
- fetched already (#62985).
-
- 05 Sep 2004; Brian Harring <ferringb@gentoo.org> portage.py ebuild.sh:
- Added use flag support to RESTRICT; usual syntax.
-
- 04 Sep 2004; Nicholas Jones <carpaski@gentoo.org> portage_data.py: Added
- FreeBSD as a BSD-type OS and merged the Darwin branch with them.
-
- 04 Sep 2004; Nicholas Jones <carpaski@gentoo.org> portage_contents.py: The
- parsing portion of a persistent contents file parser that can return the
- owner of a particular file or directory by parsing (and storing) data from
- the contents files.
-
- 04 Sep 2004; Nicholas Jones <carpaski@gentoo.org> portage_const.py: Added
- LOCALE_DATA_PATH to the constants for future gettext (internal) support.
-
- 04 Sep 2004; Nicholas Jones <carpaski@gentoo.org> output.py: Fixed the
- title changes to no clear the icon title.
-
- 04 Sep 2004; Nicholas Jones <carpaski@gentoo.org> prepman: No longer
- gzip's symlinks -- This needs to gain 'target changed' logic.
-
- 01 Sep 2004; Brian Harring <ferringb@gentoo.org> pym/portage.py:
- Cleaned up the imports, cutting down on from blah import *, instead
- importing only what is strictly needed for backards compatability.
-
- 01 Sep 2004; Brian Harring <ferringb@gentoo.org> pym/portage_exec.py:
- Nick caught this- changed setgid/setuid order so it works.
-
- 01 Sep 2004; Brian Harring <ferringb@gentoo.org> pym/portage.py:
- Added Michael Stewart's patch correcting optional args being specified
- via position- bug #61881.
-
- 01 Sep 2004; Brian Harring <ferrignb@gentoo.org> pym/portage.py,
- pym/portage_exec.py, pym/portage_checksum.py: Restructured spawn
- so that we don't have two versions; all spawn calls trace back to
- portage_exec.spawn, either through spawn_bash or spawn_sandbox.
- portage_exec.spawn is strictly an os.execve wrapper now, so bash
- doesn't have to be involved unless desired (if desired use spawn_bash).
-
- 31 Aug 2004; Brian Harring <ferringb@gentoo.org> bin/emerge: Debug
- print statement left in global scope, corrected it.
-
- 31 Aug 2004; Nicholas Jones <carpaski@gentoo.org> portage_checksum.py:
- Solar's patch to check if prelink binary exists prior to executing it.
-
- 31 Aug 2004; Nicholas Jones <carpaski@gentoo.org> portage_checksum.py:
- Only prelink-check when requested -- This is only valid for merge/unmerge
- operations -- Also happens to fix the access violations in portageq.
-
- 31 Aug 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Adjustments
- for the prelink-check changes.
-
- 31 Aug 2004; Nicholas Jones <carpaski@gentoo.org> portage_exception.py:
- Made the exceptions more hierarchial. Yes, I'm aware I can't speel.
-
- 30 Aug 2004; Brian Harring <ferringb@gentoo.org> bin/repoman: Added
- solar's patch to make the file.size check display the size of the
- offender.
-
- 30 Aug 2004; Brian Harring <ferringb@gentoo.org> pym/portage.py,
- pym/portage_util.py: Removed -all- duplicate function definitions between
- portage_util and portage. This is just -begging- for a bug where
- portage.py's definitions are fixed, but portage_util isn't.
- If you're moving code out of portage.py (good thing) please yank the def
- from portage.py. Especially if you're importing everything from new
- home of the module.
-
- 30 Aug 2004; Jason Stubbs <jstubbs@gentoo.org> pym/emerge: Added python 2.2
- compatibility fix. (#62128) Added fix to fail nicely on missing
- /etc/gentoo-release. (#62149)
-
- 30 Aug 2004; Jason Stubbs <jstubbs@gentoo.org> pym/portage.py: Fixed two
- bugs in bindbapi.aux_get() preventing retrieval of information about tbz2s.
-
- 26 Aug 2004; Brian Harring <ferringb@gentoo.org> bin/etc-update: minor
- tweak to etc-update to support single quotes in /etc/etc-update.conf
- bug (#56785). Added Mamoru Komachi (usata)'s fix for osx (#60721).
-
- 26 Aug 2004; Brian Harring <ferringb@gentoo.org> pym/portage.py,
- pym/portage_exec.py, bin/ebuild.sh: Removed the PORT_LOGDIR $0 $* | tee
- hack, and re-implemented it using spawn. Spawn has been extended so that
- fd_pipes can receive a dict of fd # -> fd, and optionally be nonblocking
- via returnpid. portage.spawn has support for a logfile optional arg- if
- specified, spawn logs stdout/stderr via tee -i -a to the specified file.
- Note portage_exec.spawn doesn't currently support this option.
-
- 25 Aug 2004; <genone@gentoo.org> pym/portage.py, +pym/portage_checksum.py,
- pym/portage_gpg.py, -pym/portage_md5.py:
- renamed portage_md5 to portage_checksum which includes support for sha1.
-
- 24 Aug 2004; Brian Harring <ferringb@gentoo.ogr> bin/ebuild.sh: Tweak to
- dyn_test to check if $S exists prior to cd'ing to it.
-
- 24 Aug 2004; Brian Harring <ferringb@gentoo.org> pym/portage.py: Fixed a
- traceback issues in dblink.treewalk when collision-protect is active.
- Also added specific check/complaint for PORT_LOGDIR='' to the config class,
- since it should either not be set, or something non-null- a null
- PORT_LOGDIR triggers a traceback in doebuild do to an attempted chmod.
-
- 23 Aug 2004; Brian Harring <ferringb@gentoo.org> pym/portage.py: Added a
- macos ranlib hack; dblink.mergeme by default resets each files mtime, which
- makes static archives merged to the fs worthless (linker notes the files
- mtime differs from an internal mtime, and bails). This closes out bug
- (#58848), and will be obsoleted when refcounts are used instead of mtime +
- md5.
-
- 21 Aug 2004; Brian Harring <ferringb@gentoo.org> bin/ebuild.sh: Expanded
- CONF_LIBDIR support, so it honors --prefix set values. (#61060)
-
- 19 Aug 2004; Marius Mauch <genone@gentoo.org> bin/repoman: Added
- FEATURES and USE to the readonly variable check.
-
- 17 Aug 2004; Brian Harring <ferringb@gentoo.org> pym/portage_data.pym:
- Adding missing imports.
-
- 17 Aug 2004; Brian Harring <ferringb@gentoo.org> bin/repoman: Minor
- tweak to make repoman detect the repolevel correctly in overlays.
- (#60298).
-
-*portage-2.0.51_pre20 (16 Aug 2004): Fixes and Public Readiness & GPG
-
- 16 Aug 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Fixed the
- imports and new modules so that the API remains constant.
-
- 16 Aug 2004; Nicholas Jones <carpaski@gentoo.org> portage_exec.py: Created
- for external operations and calls. Presently contains spawn. Mostly for
- prevention of circular imports.
-
- 16 Aug 2004; Nicholas Jones <carpaski@gentoo.org> portage_md5.py: using
- the spawn call from portage_exec.py to avoid the circular import.
-
- 16 Aug 2004; Brian Harring <ferringb@gentoo.org> bin/repoman: Fixed a
- traceback related to file.size and --fix, added compatability tweaks for
- xmllint.
-
- 16 Aug 2004; Nicholas Jones <carpaski@gentoo.org> emerge: Fixed the lock
- handling and unique_array calls.
-
-*portage-2.0.51_pre19 (16 Aug 2004): Fixes and Public Readiness & GPG
-
- 16 Aug 2004; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Add predict
- for gpg verification.
-
- 16 Aug 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Fixed the
- lock code to use the external module.
-
- 16 Aug 2004; Nicholas Jones <carpaski@gentoo.org> portage_gpg.py: Fixed
- the writing operations and access violations.
-
- 16 Aug 2004; Nicholas Jones <carpaski@gentoo.org> portage_locks.py: Fixed
- the code so it acutally works and is used.
-
- 16 Aug 2004; Nicholas Jones <carpaski@gentoo.org> portage_md5.py: Fixed
- the lock calls.
-
-*portage-2.0.51_pre18 (16 Aug 2004): Fixes and Public Readiness & GPG
-
- 16 Aug 2004; Nicholas Jones <carpaski@gentoo.org> dolib*: Added LV's
- CONF_LIBDIR patch to help out the 32/64 bit lib migrations.
-
- 16 Aug 2004; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: CONF_LIBDIR
- patch updates. Added a possible fix for the export issues in environment.
-
- 16 Aug 2004; Nicholas Jones <carpaski@gentoo.org> prepall/lib: CONF_LIBDIR
-
- 16 Aug 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Adjusted the
- manifest/gpg code to reduce the output on missing sigs. Changed the GPG
- homedir to the PORTAGE_GPG_DIR instead of using rsync for the keyring --
- This requires manual intervention.
-
- 16 Aug 2004; Nicholas Jones <carpaski@gentoo.org> sandbox: Added the 32/64
- paths in for 32/64 lib migrations.
-
- 16 Aug 2004; Brian Harring <ferringb@gentoo.org> pym/cvstree.py
- pym/portage.py: Round #2 of ignorecvs, now w/ sane regex goodness and an
- addition to digest(gen|check) to use the same cvs filter for Manifests.
- (#46070).
-
- 15 Aug 2004; Nicholas Jones <carpaski@gentoo.org> emerge: message fix for
- packages.provided.
-
- 15 Aug 2004; Nicholas Jones <carpaski@gentoo.org> repoman: Made file.size
- a warning for the time being.
-
- 15 Aug 2004; Nicholas Jones <carpaski@gentoo.org> cvstree.py: Removed the
- auto-ignore regex as it is broken AND it breaks Manifests due to excess
- files allowed into them.
-
- 15 Aug 2004; Nicholas Jones <carpaski@gentoo.org> emerge: A couple changes
- to the select_dep exception handling for the signing code.
-
- 15 Aug 2004; Nicholas Jones <carpaski@gentoo.org> env-update.sh: Added
- Spanky's env-update shell script version.
-
- 15 Aug 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Moved more
- code into seperate modules. Added support for Manifest verification and
- usage of 'gpg' 'strict' 'severe' to enable various condition responses.
-
- 15 Aug 2004; Nicholas Jones <carpaski@gentoo.org> portage_const.py: Moved
- all constants to this module. (All uppercase defines)
-
- 15 Aug 2004; Nicholas Jones <carpaski@gentoo.org> portage_data.py: Contains
- all calculated information. uid/gid info. system-specific values. All probed
- information should go here.
-
- 15 Aug 2004; Nicholas Jones <carpaski@gentoo.org> portage_exception.py:
- Added many exceptions for the GPG verification code. Added many general
- exceptions to help differentiate from explicit portage exceptions and
- those issued by python.
-
- 15 Aug 2004; Nicholas Jones <carpaski@gentoo.org> portage_gpg.py: Handling
- of gpg verification code and keyring management/trust.
-
- 15 Aug 2004; Nicholas Jones <carpaski@gentoo.org> portage_locks.py: Moved
- the lock code out of portage.py.
-
- 15 Aug 2004; Nicholas Jones <carpaski@gentoo.org> portage_md5.py: Moved the
- MD5 calculation code out of portage.py.
-
- 15 Aug 2004; Nicholas Jones <carpaski@gentoo.org> portage_util.py: Moved
- writemsg and unique_array into portage_util.
-
- 13 Aug 2004; Brian Harring <ferringb@gentoo.org> bin/ebuild.sh: Added
- support for ECONF_LIBDIR; if it isn't defined, then --libdir isn't
- passed to the configure script. If it is defined, then the configure
- script gets --libdir=/usr/${ECONF_LIBDIR}.
-
- 13 Aug 2004; Brian Harring <ferringb@gentoo.org> pym/cvstree.py:
- Added robbat2's patch to ignore files that cvs ignores. (#46070).
-
- 13 Aug 2004; Brian Harring <ferringb@gentoo.org> bin/ebuild.sh: Shifted
- the add* sandbox function definitions to before profile.bashrc srcing.
- As bug #60147 demonstrated, profiles occasionally need to adjust
- SANDBOX_WRITE (current case being for /usr/lib64/{conftest,cf}).
-
- 13 Aug 2004; Brian Harring <ferringb@gentoo.org> bin/emerge: Related to
- bug #60256, adjusted format_size so that is always returns a string.
-
- 13 Aug 2004; Jason Stubbs <jstubbs@gentoo.org> bin/emerge:
- Relocated blocker checking code to before pkgsettings.setcpv() is called
- on it in depgraph.create()
-
- 13 Aug 2004; Brian Harring <ferringb@gentoo.org> pym/portage.py:
- Tweaked lockfile, so it attempts a non-blocking lock_ex first, then
- states it's waiting on lock blar, then attempts a blocking lock. This
- will be useful for informing the user why portage seems to have hung,
- and good for debugging.
-
- 13 Aug 2004; Brian Harring <ferringb@gentoo.org> bin/emerge-webrsync:
- Added support for snapshot md5sum's (mirrors now carry them).
- This is used to ensure the fetched snapshot is sane; if it's sane,
- then we reuse it for sync'ing. This nulls the -n option, so it's been
- removed. Closes out #15990, but no longer automatically forcing a refetch.
- Refetches are only forced if the md5 is invalid.
-
- 12 Aug 2004; Brian Harring <ferringb@gentoo.org> pym/portage.py:
- Cleaned up fetch a bit more, saner error messages when unable to
- write to DISTDIR.
-
- 12 Aug 2004; Jason Stubbs <jstubbs@gentoo.org> bin/emerge:
- Fixed the fix for the earlier traceback on installed package not being
- in PORTDIR to remove duplicate work.
-
- 11 Aug 2004; Brian Harring <ferringb@gentoo.org> bin/ebuild.sh:
- Added FEATURES="autoconfig" support. (#55476)
-
- 11 Aug 2004; Brian Harring <ferringb@gentoo.org> pym/portage.py:
- Added check to lockfile and fetch, if file already is owned by
- portage group, don't try and chown it. This will close #60079.
-
- 10 Aug 2004; Brian Harring <ferringb@gentoo.org> bin/emerge: Fixed a
- traceback issue, an ebuild is no longer in the tree, but is installed
- and needs to be used in the depgraph- the problem was, emerge
- assumed the ebuild was in porttree's db, when vartree should be used
- if the package is known to be installed.
-
- 10 Aug 2004; Brian Harring <ferringb@gentoo.org> pym/portage.py:
- Killed a couple of corner cases for non-root fetch and lockfile calls.
-
- 10 Aug 2004; Brian Harring <ferringb@gentoo.org> pym/portage.py
- bin/ebuild.sh: Add check to ensure that install phase has been
- ran prior to qmerge phase being attempted. This can happen
- when the user is using ebuild to step through the phases.
- Corrected bug in listdir where it would return None, rather then
- []- all callee's expect a returned list, not None. Same for ftype,
- cause's a tb if you just haphazardly rely on cachedir's return.
-
- 09 Aug 2004; Brian Harring <ferringb@gentoo.org> pym/portage.py:
- Leave /var/tmp with it's own permissions, chowning/chmoding just
- /var/tmp/portage (#37521).
-
- 09 Aug 2004; Brian Harring <ferringb@gentoo.org> bin/ebuild.sh
- pym/portage.py: Corrected bug in profile.bashrc support, added aliases
- for saving/restoring IFS (remove_path_entry and profile.bashrc loop
- adjust IFS temporarily). Closes #59749.
-
- 08 Aug 2004; Brian Harring <ferringb@gentoo.org> dispatch-conf:
- Converted os.rename calls to shutil.move; the former can't cross fs's,
- the latter can. (#46148)
-
- 07 Aug 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Fixed the
- cache updates so that an env-update forces ld updating.
-
- 07 Aug 2004; Jason Stubbs <jstubbs@gentoo.org> pym/portage.py
- pym/portage_dep.py: Added a new parameter to use_reduce so that !arch?
- checks can be adhered to even when matchall=1 and arch is not is masklist.
-
- 07 Aug 2004; Jason Stubbs <jstubbs@gentoo.org> pym/portage.py
- pym/portage_dep.py: Added profile masked use flags to repoman check.
- Moved || refactoring to a separate function. Reworked use_reduce logic
- into simpler sections.
-
- 05 Aug 2004; Brian Harring <ferringb@gentoo.org> pym/portage_dep.py:
- What can I say, I can't get enough of bug (#59574).
- I'm operating under the assumption there still is a bug in the
- use_reduce logic, soo I've left a fairly massive amount of debugging
- code in place that's currently disabled. It's *very* useful for
- tracking exactly what/how use_reduce decides on a operator node.
-
- 05 Aug 2004; Brian Harring <ferringb@gentoo.org> pym/portage.py:
- I introduced a bug in the previous commit, basically
- portage.settings.archlist list of arch keywords was being actively
- pruned by repoman as it stepped through arches for dep checking.
- Basically, needed to make a copy of archlist rather then using the
- actual archlist object. Should be the final issue for (#59574).
-
- 05 Aug 2004; Brian Harring <ferringb@gentoo.org> pym/portage.py: Fix
- in the same area, if use="all" (repoman wants -every use flag- checked),
- it should call use_reduce w/ matchall set appropriately. (#59574)
-
- 05 Aug 2004; Brian Harring <ferringb@gentoo.org> pym/portage.py: And...
- repoman's arch check got hosed again. The code in dep_check for
- building a masklist was incorrect, 3 line fix. (#59574)
-
-*portage-2.0.51_pre17 (03 Aug 2004): Fixes and Public Readiness
-
- 03 Aug 2004; Nicholas Jones <carpaski@gentoo.org> portage_dep.py: Again
- fixed the || exceptions to understand nested legal || statements.
-
-*portage-2.0.51_pre16 (03 Aug 2004): Fixes and Public Readiness
-
- 04 Aug 2004; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Changed the
- virtual/glibc references to virtual/libc.
-
- 04 Aug 2004; Nicholas Jones <carpaski@gentoo.org> prepman: No longer runs
- gzip on .keep files.
-
- 04 Aug 2004; Nicholas Jones <carpaski@gentoo.org> getbinpkg.py: Added in
- an active/passive option for FTP connections -- Requires the appending of
- an asterisk to the HOST portion of the ftp string to use active connections.
-
- 04 Aug 2004; Nicholas Jones <carpaski@gentoo.org> make.conf*: Updates for
- the active-connection ftp option.
-
- 04 Aug 2004; Nicholas Jones <carpaski@gentoo.org> emergehelp.py: Removed
- the bin/ version and replaced it with a duplicated pym/ version.
-
- 03 Aug 2004; Brian Harring <ferringb@gentoo.org> pym/portage.py:
- Added check to ensure /var/tmp/portage's permissions were sane. (#56665)
-
- 03 Aug 2004; Brian Harring <ferringb@gentoo.org> bin/repoman: Added
- explicit check for missing files directory, rather then ignoring it.
- Also added an explicit commit-time check for CVS/Entries being sane.
- (#57141).
-
- 03 Aug 2004; Brian Harring <ferringb@gentoo.org> bin/repoman: reworked
- the qacats definition so that the type of bug that borked emerge help is
- no longer possible.
-
- 03 Aug 2004; Brian Harring <ferringb@gentoo.org> pym/portage.py,
- bin/emerge: Corrected emerge --fetch-all-uri issues, now works. Basically,
- FEATURES="cvs" emerge -f blar == emerge --fetch-all-uri blar. The fetch
- option handling in emerge could use some cleanup.
-
- 03 Aug 2004; Brian Harring <ferringb@gentoo.org> pym/portage.py: Ongoing
- cleanup in emerge -fp; this corrects a minor naggle affecting previous
- releases, where emerge -fp would still attempt to do md5 checks on files.
-
- 03 Aug 2004; Brian Harring <ferringb@gentoo.org> pym/portage.py: Corrected
- logic for fetch(..., ..., use_locks=1,listonly=1) attempting to use locks,
- when fetch is just printing the src_uri's. (#59394).
-
- 03 Aug 2004; Brian Harring <ferringb@gentoo.org> bin/repoman: Corrected the
- borkage I introduced into repoman's help option- it was throwing a traceback
- due to file.executable's key name being typoed.
-
- 03 Aug 2004; Nicholas Jones <carpaski@gentoo.org> portage_dep.py: Enhanced
- the invalid depend string identification for || (and &&) strings.
-
-*portage-2.0.51_pre15 (03 Aug 2003): Fixes and Public Readiness
-
- 03 Aug 2004; Brian Harring <ferringb@gentoo.org> bin/repoman: Corrected
- typo, trying to remove .backup_metadata.dtd rather then metadata.dtd.
-
- 03 Aug 2004; Nicholas Jones <carpaski@gentoo.org> portage_dep.py: Changing
- from dep_opconvert to use_reduce introduced a bug due to how OR'd lists are
- managed by the remaining dep handling functions -- Fixed by emulating the
- format in use_reduce.
-
- 02 Aug 2004; Brian Harring <ferringb@gentoo.org> bin/repoman, pym/portage.py:
- Added local caching of metadata.dtd, to prevent flaky connections from
- flagging a packages metadata.xml as invalid due to xmllint failing to fetch
- metadata.dtd. Simplified version of patch in (#57210).
-
- 02 Aug 2004; Brian Harring <ferringb@gentoo.org> bin/ebuild.sh,
- pym/portage.py: Added support for src'ing a profile bashrc. Fex, if
- /etc/make.profile/profile.bashrc exists, it is sourced prior to ebuild.sh
- defining any of it's functions. (#58415).
-
- 02 Aug 2004; Brian Harring <ferringb@gentoo.org> bin/emerge-webrsync: General
- cleanup; uses the make.conf defined FETCHCOMMAND for fetching (fixing #57887),
- runs emerge metadata after a successful sync also.
-
-*portage-2.0.51_pre14 (02 Aug 2003): Fixes and Public Readiness
-
- 02 Aug 2004; Nicholas Jones <carpaski@gentoo.org> do*: Added exit calls
- on failures and changed the install to use short options for BSD compat.
-
- 02 Aug 2004; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: unset
- GLOBIGNORE added. Echo out the confingure command from econf. Moved the
- maketest code to dyn_preinst so it didn't force-run maketest. Added a
- patch for keyword expansion (requires portage-2.0.51 for use).
-
- 02 Aug 2004; Nicholas Jones <carpaski@gentoo.org> emerge: --fetch-all-uri
- added to force all URIs to be downloaded (works like features=cvs). libc
- version printing enhancements. Fixed the binary package selection in an
- alt ROOT. Headers and libtool added to info. --ask removed on a resume.
-
- 02 Aug 2004; Nicholas Jones <carpaski@gentoo.org> prepstrip: Change ewarn
- to echos to stderr with beeps.
-
- 02 Aug 2004; Nicholas Jones <carpaski@gentoo.org> repoman: Added to the
- OK message output for issues that fail.
-
- 02 Aug 2004; Nicholas Jones <carpaski@gentoo.org> make.conf: Typo correction
- patch added.
-
- 02 Aug 2004; Nicholas Jones <carpaski@gentoo.org> output.py: Added kterm
- to the titlebar terminals.
-
- 02 Aug 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Added a message
- regarding the location of the virtuals file (move to /etc/portage/profile).
- Added a invalid-mirror message and a missing URI message for fetching. Added
- the selinux sandbox patch. Removed the old dep_opconvert call as all ?:
- syntax is gone.
-
- 02 Aug 2004; Nicholas Jones <carpaski@gentoo.org> portage_dep.py: Fixed the
- use_reduce code to properly handle negative requirements on masked flags.
-
- 02 Aug 2004; Nicholas Jones <carpaski@gentoo.org> tbz2tool.c: Made all the
- comments to C style.
-
- 02 Aug 2004; Nicholas Jones <carpaski@gentoo.org> libsandbox: Code from
- Seth Robertson that tracked down all adjuct flags for read operations that
- did not invoke a write operation.
-
- 01 Aug 2004; Masatomo Nakano <nakano@gentoo.org> bin/emerge: Added
- message about ._cfg* files after emerge sync.
-
- 01 Aug 2004; Masatomo Nakano <nakano@gentoo.org> pym/portage.py: Modified
- updating /etc/portage/package.* logic to see CONFIG_PROTECT and
- CONFIG_PROTECT_MASK.
-
- 01 Aug 2004; Marius Mauch <genone@gentoo.org> bin/emerge:
- Added a warning for `emerge /path/to/ebuild`
-
- 01 Aug 2004; Marius Mauch <genone@gentoo.org> bin/emerge:
- Fixing broken logic for the `emerge rsync` deprecation notice (it only showed
- up on `emerge --rsync`).
-
- 01 Aug 2004; Brian Harring <ferringb@gentoo.org> pym/portage.py bin/emerge:
- Fixed the lockfile/unlockfile functions so that they correctly support having
- an int passed in (the fd #), or having a file object passed in. Corrected
- lockfile so that the lock is owned by portage group, with g+w permissions
- (this is needed since sudo emerge blar can bail, leaving a stale lock that
- non-root usage cannot remove). Added locking to emerge's emergelog function,
- preventing log messages from potentially getting mixed together. Added check
- to fetch function to complain if unable to write to DISTDIR (previously the
- fetcher just bailed, stepping through each src_uri). Added lockfiles for
- fetching/md5ing of the src- these lockfiles are stored in a subdirectory
- (locks_in_subdir=".locks") if specified, and locking is controlled via
- use_locks (defaults to on). emerge -f no longer requires root/sudo to run
- (ebuild never had this restriction). (#42969)
-
- 31 Jul 2004; Jason Stubbs <jstubbs@gentoo.org> pym/portage.py bin/emerge:
- Removed USE-based SLOT support. Removed uselist from getslot methods.
-
- 29 Jul 2004; Brian Harring <ferringb@gentoo.org> bin/repoman: Added check for
- files over 20k in a packages files directory. file.size, repoman treats it
- as a a failure. Added the repoman manpage entries for file.size, and
- file.executable (I missed file.executable earlier).
-
- 28 Jul 2004; Masatomo Nakano <nakano@gentoo.org> pym/portage.py: Fixed a bug
- emerge doesn't pkg_setup() with -k/-K option. (#25152)
-
- 28 Jul 2004; Masatomo Nakano <nakano@gentoo.org> pym/portage.py:
- Added ._cfg* file for updating /etc/portage/packages.* during global
- update.
-
- 28 Jul 2004; Masatomo Nakano <nakano@gentoo.org> bin/emergehelp.py,
- man/emerge.1: Added explanation of --newuse to manpage/help.
- Added information of --verbose to manpage.
-
- 27 Jul 2004; Jason Stubbs <jstubbs@gentoo.org> pym/portage.py:
- Reversed the order of profile virtuals (dirVirtuals) so that a cascading
- profile's virtuals are stacked in the correct order.
-
- 26 Jul 2004; Brian Harring <ferringb@gentoo.org> bin/emerge:
- Removed the `which blah` calls- A) stage1 lacks which, B) which searches
- $PATH- all sane shells do this already when handed a command that
- isn't absolute path. These which calls were used in
- commands.getstatusoutput() calls, which starts up a shell with
- the arg passed to the shell, so "`which blar` args" isn't needed.
-
- 26 Jul 2004; Brian Harring <ferringb@gentoo.org> bin/repoman:
- Added file.executable check- ebuilds, digests, Manifest, ChangeLog, and
- metadata.xml don't need the executable bit set. CVS preserves it upon
- commit, so we do the check prior to commit. (#55647)
-
- 26 Jul 2004; Brian Harring <ferringb@gentoo.org> bin/emerge:
- Removed the hardcoding of uname for emerge info, using which to find it
- instead.
-
- 26 Jul 2004; Brian Harring <ferringb@gentoo.org> bin/repoman:
- Added check to ensure manifest recommit is at least possible when
- committing, and corrected handling of CVS/Root files for OSX machines.
- Removed readline import, doesn't seem to be used at all.
-
- 26 Jul 2004; Jason Stubbs <jstubbs@gentoo.org> bin/emerge pym/portage.py:
- Refactored slot code into portage.py to remove usage off portage_dep.
-
- 25 Jul 2004; Brian Harring <ferringb@gentoo.org> bin/dohtml:
- Corrected a bug in dohtml where it was unable to install files with a space
- in their name (#58258)
-
- 25 Jul 2004; Jason Stubbs <jstubbs@gentoo.org> bin/emerge pym/portage.py:
- Deprecated --inject and added support for package.provided in both the
- profiles and /etc/portage/profile directory.
-
- 24 Jul 2004; Jason Stubbs <jstubbs@gentoo.org> bin/emerge pym/portage.py:
- Added USE flag based SLOT support.
-
- 24 Jul 2004; Jason Stubbs <jstubbs@gentoo.org> bin/emerge pym/portage.py:
- Added USE flag based PROVIDE support. (#32114)
-
- 24 Jul 2004; Jason Stubbs <jstubbs@gentoo.org> src/sandbox-1.1/libsandbox.c:
- Fixed lchown sandbox bug. (#58084)
-
- 23 Jul 2004; Brian Harring <ferringb@gentoo.org> bin/repoman: Corrected a
- false positive on the ebuild.nesteddie check. Basically it wasn't looking
- to see if the line was active or not (fex # (die), bash skips, but repoman
- caught). Aside from that, that check is still capable of missing multiline
- ebuild.nesteddie instances. (#33011)
-
- 22 Jul 2004; Brian Harring <ferringb@gentoo.org> bin/ebuild.sh: Re-enabled
- binary QA interceptors, with a minor twist- the QA_INTERCEPTORS is used to
- specify what QA interceptors will be defined. These functions are no longer
- saved in the ebuild's stored env (exist *only* in depend phase), and have been
- expanded to identify if it's the ebuild, or eclass that's triggering the QA
- Notice (#54652). The interceptors behaviour when the binary is missing has
- been corrected to correctly output "missing command $bin: args".
-
- 21 Jul 2004; Masatomo Nakano <nakano@gentoo.org> bin/emerge: Fixed bug
- which blocks a package itself with -U option.
-
- 20 Jul 2004; Masatomo Nakano <nakano@gentoo.org> bin/repoman: Fixed
- profile cache problem. (#43601, #56170)
-
- 20 Jul 2004; Marius Mauch <genone@gentoo.org> bin/ebuild.sh:
- added usev() and hasv() as complement to useq() and hasq().
-
- 18 Jul 2004; Brian Harring <ferringb@gentoo.org> bin/repoman: Corrected
- ebuild.allmasked check, so that it checks for an available version across all
- arches, rather then the last arch processed. Typo fixed also- bugs #57356 and
- #57068.
-
- 17 Jul 2004; Jason Stubbs <jstubbs@gentoo.org> bin/repoman: Added patch from
- #49126 that makes repoman check that a USE flag from use.local.desc applies
- to the packages that make use of it.
-
- 14 Jul 2004; Marius Mauch <genone@gentoo.org> bin/emerge:
- added a deprecation warning for --upgradeonly
-
- 14 Jul 2004; Marius Mauch <genone@gentoo.org> bin/etc-update:
- Added a hint to etc-update so people that don't know what to do don't use -3
- or -5 by accident.
-
- 12 Jul 2004; Jason Stubbs <jstubbs@gentoo.org> bin/emerge:
- Adjusted masked binary checking code to exclude --usepkgonly.
-
- 10 Jul 2004; Jason Stubbs <jstubbs@gentoo.org> man/portage.5:
- Adjusted documentation for user use.mask to match implementation.
-
-*portage-2.0.51_pre13 (09 Jul 2004): Fixes and Public Readiness
-
- 21 Jun 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Included
- ferringb's logic fix for the repoman code. Fix that prevents vardb from
- using all files and directories in the vardb as keys instead of only the
- proper ones. Code touchups.
-
- 21 Jun 2004; Nicholas Jones <carpaski@gentoo.org> prepstrip: Fixes that
- result in BSD compat and split the regex -- find doesn't do extended.
-
- 21 Jun 2004; Nicholas Jones <carpaski@gentoo.org> emerge: included a uname
- fix for bsd. Added --fetch-all-uri as a way to get all URIs downloaded for
- a package regardless of conditionals.
-
- 21 Jun 2004; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Moved the
- declaration of ebuild_phase toward the top.
-
- 09 Jul 2004; Jason Stubbs <jstubbs@gentoo.org> bin/emerge:
- Fixed incorrect assignment of depgraph.mydbapi[] objects.
-
- 07 Jul 2004; Marius Mauch <genone@gentoo.org> bin/repoman:
- Solved a big memory problem in repoman where a full scan required several
- gigabytes, caused by apparently unused objects.
-
- 04 Jul 2004; Jason Stubbs <jstubbs@gentoo.org> bin/emerge pym/portage.py:
- Fixed issue that allowed installed virtuals to overide user virtuals.
-
- 03 Jul 2004; Jason Stubbs <jstubbs@gentoo.org> bin/emerge: Added masked
- package removal to binpkg candidates before getting the best. (#55871)
-
- 01 Jul 2004; Jason Stubbs <jstubbs@gentoo.org> pym/portage.py:
- Fixed exception catching on module import.
-
- 27 Jun 2004; Jason Stubbs <jstubbs@gentoo.org> bin/repoman:
- Fixed year (2003 -> 2004) in copyright lines.
-
- 27 Jun 2004; Marius Mauch <genone@gentoo.org> pym/portage.py:
- Added a check for symlinked dirs to the collision-protect code so it doesn't
- double-check files in symlinked dirs with wrong pathnames.
-
- 26 Jun 2004; Masatomo Nakano <nakano@gentoo.org> bin/regenworld: Fixed
- regenworld. It always failed without -h or --help.
-
- 26 Jun 2004; Masatomo Nakano <nakano@gentoo.org> bin/emerge: Added more
- messages when emerge fails due to masked package.
-
- 26 Jun 2004; Jason Stubbs <jstubbs@gentoo.org> portage.py:
- Added pkgsplit check on values returned by vardbapi.cpv_all()
-
- 26 Jun 2004; Jason Stubbs <jstubbs@gentoo.org> repoman: Removed a stray dot.
-
- 25 Jun 2004; Jason Stubbs <jstubbs@gentoo.org> repoman: Changed CVS
- header regex to "Gentoo Foundation"
-
-*portage-2.0.51_pre12 (21 Jun 2003): Fixes
-
- 21 Jun 2004; Nicholas Jones <carpaski@gentoo.org> emerge: Removed the
- extra spawn of portageq as sandbox handles the python pyo's accesses now.
-
- 21 Jun 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Fixes for
- the HOME issues. Correction of permissions from 6770 to 2770.
-
- 21 Jun 2004; Masatomo Nakano <nakano@gentoo.org> pym/portage.py:
- Fixed problem that 'emerge something' would install all depeneded pkgs
- even if they are already installed.
-
-*portage-2.0.51_pre11 (21 Jun 2004): Fixes
-
- 21 Jun 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Disabled the
- URL encoding for BINHOST as it will take a lot of work to get it right.
- Execve try/except added to handle missing binaries and such. Changed the
- default HOME to be TMPDIR/homedir. Try/except on db close operations. Add
- in missing unlocks. Hacked in the /var/lib/portage in a chroot fix.
-
- 21 Jun 2004; Nicholas Jones <carpaski@gentoo.org> emerge: Added more info
- into the profile for the stacked profiles.
-
- 21 Jun 2004; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: The userland
- fixes for *BSD. QA Notice fix for IUSE verbosity on global scope stuff.
- Missing quote fix for QA intercepters. Return 0 in the use_* functions.
- Removed AA from the readonly list.
-
- 21 Jun 2004; Nicholas Jones <carpaski@gentoo.org> dohtml: Karl's update
- which is mostly a rewrite.
-
- 21 Jun 2004; Jason Stubbs <jstubbs@gentoo.org> bin/emerge:
- Modified emerge to build up a fakedb regardless of 'empty' in params,
- and to incrementally add any packages added to the dep graph.
- (#1343, #8810, #54608)
-
- 20 Jun 2004; Masatomo Nakano <nakano@gentoo.org> bin/repoman:
- Fixed problem which fails to detect IUSE value. (#21544)
-
- 20 Jun 2004; Masatomo Nakano <nakano@gentoo.org> bin/emerge:
- Added an exception code for broken timestamp.chk. (#54380)
-
- 20 Jun 2004; Jason Stubbs <jstubbs@gentoo.org> pym/ebuild.sh: Adjusted
- returns for use_with and use_enable to only return 1 on invalid usage.
-
- 20 Jun 2004; Masatomo Nakano <nakano@gentoo.org> pym/portage.py:
- Fixed problems that portage doesn't block with --update option(#52377)
- and fail to block virual packages(#52506).
-
- 20 Jun 2004; Masatomo Nakano <nakano@gentoo.org> pym/portage.py:
- Fixed CONFIG_PROTECT/CONFIG_PROTECT_MASK in /etc/csv.env(#51646).
-
- 20 Jun 2004; Masatomo Nakano <nakano@gentoo.org> bin/ebuild.sh:
- Added error message when pkg_config is not defined. (#51167)
-
- 14 Jun 2004; Jason Stubbs <jstubbs@gentoo.org> pym/portage.py:
- Fixed bug in portdbapi.aux_get where a problem ebuild would cause a lock
- to not be released.
-
- 09 Jun 2004; Marius Mauch <genone@gentoo.org> bin/emerge:
- Changed emerge --info to reflect libc versions other than glibc.
-
- 08 Jun 2004; Marius Mauch <genone@gentoo.org> pym/portage.py:
- Added the collision-protect feature that prevents packages from
- overwriting files they don't own. Has to be enabled with
- FEATURES=collision-protect as it needs more testing before it
- can be enabled by default (bug #28228).
-
- 05 Jun 2004; Marius Mauch <genone@gentoo.org> bin/ebuild.sh:
- Fix rpm support by changing rpm to rpmbuild (bug #13508).
-
- 05 Jun 2004; Marius Mauch <genone@gentoo.org> bin/regenworld:
- Added a --help message ro regenworld (bug #37539).
-
- 03 Jun 2004; Jason Stubbs <jstubbs@gentoo.org> portage.py: Fixed bug #52720
- and restored patch for a 35% drop in dep calc time.
-
-*portage-2.0.51_pre10 (02 Jun 2003): Fixes
-
- 02 Jun 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Added MANPATH
- to the colon_seperated list -- parsing code NEEDS TO BE FIXED. * and ~*
- matches allowed for package.use for arch-development -- ~* implies *.
-
- 02 Jun 2004; Nicholas Jones <carpaski@gentoo.org> emerge: Fix for a new
- typo in the logger function that prevented emerge.log from being written.
-
- 02 Jun 2004; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: make test
- only occurs when enabled in FEATURES and not in RESTRICT -- defaults OFF.
- Added ferringb's local B_* fix for eclasses. Made db vars readonly via bash
- for non-depend phases after the global scope.
-
- 02 Jun 2004; Jason Stubbs <jstubbs@gentoo.org> pym/portage.py: Reverted
- adjusted made to config.setcpv on 18 May 2004 due to bug #52720.
-
- 01 Jun 2004; Marius Mauch <genone@gentoo.org> bin/repoman,bin/regenworld:
- only messages about broken log entries in regenworld when called with
- --debug. Fix signing stuff in repoman.
-
-*portage-2.0.51_pre9 (22 May 2003): Speedups and bug fixes
-
- 22 May 2004; Nicholas Jones <carpaski@gentoo.org> portage_db_template.py:
- Added in last-three caching for db modules. Ensuring that keys are strings.
-
- 22 May 2004; Nicholas Jones <carpaski@gentoo.org> portage_db_cpickle.py: Fix
- that ensures the Unpickler works properly.
-
- 22 May 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Updates for
- environment files ordering in ebuild.sh. Added in normalize_path to handle
- the really annoying "'//' is a legal prefix" issue with os.path.normpath.
- Applied normalize_path to cacheddir() so that the keys match and we have
- more cache hits and better performance. Added statistics to to cacheddir
- available with noise=2. Do not return a blocker when doing a zerolist.
- Fixed the stacking functions to properly order the incrementals and apply
- the removals forward. Changed all string.atoi() to int(). Added in the
- selinux changes for spawning the fetch command. Added a patch that allows
- ebuild.sh to display 'validcommands'. close_caches() added to allow atexit
- to close all the DB connections. close_portdbapi_cache() handles the global
- that has all db instances -- This avoids problems with the API nulling the
- internal links. Adjusted the db handling for sync() calls and removed some
- object duplication db calls.
-
- 22 May 2004; Nicholas Jones <carpaski@gentoo.org> getbinpkg.py: Fix for HTTP
- redirects (301 and 302) so that the redirect opens a new connection to the
- move-location server -- This allows us to redirect to a different machine.
-
- 22 May 2004; Nicholas Jones <carpaski@gentoo.org> emerge: For iuse output we
- now ensure that the correct db is referenced. The worldfile additions do not
- occur when a packages is new and updates are not being performed. Patch to
- add binutils to the emerge info.
-
- 22 May 2004; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Rearranged the
- sourcing of the profiles and the special environment files. Added src_test
- as an optional test method -- Some packages are extremely dumb and need to
- be prevented from using this via RESTRICT.
-
- 18 May 2004; Jason Stubbs <jstubbs@gentoo.org> portage.py: Minor speed
- improvement in config.setcpv preventing a useless regenerate
-
- 18 May 2004; Jason Stubbs <jstubbs@gentoo.org> portage.py: Reversed the
- order of config.getvirtuals' stack_dictlist call for correct ordering.
-
-*portage-2.0.51_pre8 (16 May 2003): Big cleanups & Ebuild QA Stuff.
-
- 16 May 2004; Nicholas Jones <carpaski@gentoo.org> portage_exception.py: New
- file containing portage exceptions. Added 'CorruptionError' for the db code.
-
- 16 May 2004; Nicholas Jones <carpaski@gentoo.org> portage_util.py: New file
- that contains utility functions. Allows db_modules to use portage functions
- without a circular dependency. CODE IS DUPLICATED and needs to be fixed in
- portage.py to use this new module. grab*, getconfig, varexpand, pickle_*,
- ReadOnlyConfig
-
- 16 May 2004; Nicholas Jones <carpaski@gentoo.org> portage_db_*: Modules
- do not have an __init__ any longer -- module_init() is called by the
- template after loading the config and setting the 5 variables inside of
- the class. self.config is a ReadOnlyConfig object which is just a dict
- that you can't write to. Throws CorruptionError when the read functions
- return exceptions.
-
- 16 May 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Reversed the
- virtuals order so it matches the documented format. When an ebuild does not
- exist, we now raise a KeyError with a useful string. 'couple minutes' changed
- to 'couple of minutes' for Seemant. ;)
-
- 16 May 2004; Nicholas Jones <carpaski@gentoo.org> prepstrip, ebuild.5,
- repoman.1: Typo corrections/text replacements.
-
- 16 May 2004; Nicholas Jones <carpaski@gentoo.org> emerge: Added deprecation
- notice to 'emerge rsync' to favor 'emerge sync'. Modified the emerge.log
- code to not change permissions on the files beyond what is necessary -- NEEDS
- TO BECOME AN ADDITIVE FUNCTION. Produce a warning instead of a traceback
- when an ebuild does not exist for -U.
-
- 16 May 2004; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Touched up the
- logging stuff at the top. Removed extra_functions.sh. src_compile additions
- so that it can run make without having a config file requirement.
-
- 16 May 2004; Nicholas Jones <carpaski@gentoo.org> dobin,dosbin: Removed the
- duplicated stripping from the helper tools.
-
- 09 May 2004; Jason Stubbs <jstubbs@gentoo.org> pym/portage_dep.py:
- fixed bug in use_reduce that caused returned list to be flattened.
-
- 30 Apr 2004; Marius Mauch <genone@gentoo.org> bin/ebuild.sh:
- modified ebuild.sh error message on seemants and roger55s request.
-
-*portage-2.0.51_pre7 (26 Apr 2003): Big cleanups & Ebuild QA Stuff.
-
- 25 Apr 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Fixed masking
- problems due to the new stacking functions and empty-value stripping.
-
-*portage-2.0.51_pre6 (25 Apr 2003): Big cleanups & Ebuild QA Stuff.
-
- 25 Apr 2004; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Dropped selinux
- from the IUSE complainer list.
-
- 25 Apr 2004; Nicholas Jones <carpaski@gentoo.org> repoman: Made it use the
- constants defined in portage.py for the world file.
-
- 25 Apr 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Modifications
- so that the virtuals are incremented in the proper order and reversed the
- final values so that the first virtual is the 'best' one. Try to put the
- INHERITED variable back into the environment before calling out to portage.
- This makes the ECLASS QA notices actually valid.
-
-*portage-2.0.51_pre5 (25 Apr 2003): Big cleanups & Ebuild QA Stuff.
-
- 25 Apr 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Fixed name
- of the world file constant. Fixed the virtuals loading and vartree creation
- circular dep with a repitition hack. Fixed the virtuals loading function so
- that it doesn't destroy the virtuals before saving them.
-
- 25 Apr 2004; Nicholas Jones <carpaski@gentoo.org> emerge: Fixed all the
- references to the world file to use portage constants. Made the verbose
- output for use flags call unique array to make sure values aren't duped.
-
- 25 Apr 2004; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Typo fix for
- inherited, and removed the delay.
-
-*portage-2.0.51_pre4 (25 Apr 2003): Big cleanups & Ebuild QA Stuff.
-
- 25 Apr 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Creation of
- a great number of 'constants' that were hard coded in various places and
- now are created in a cascaded style, for the most part. Exception handling
- for the loading of modules added along with verbose messages about the
- exception being handled for user info. PRIVATE_PATH is a new constant that
- points to the new directory we will be using for portage internal data --
- It is a secure directory that only group portage may write, read, and scan.
- PORTAGE_CACHEDIR is deprecated -- The real cache directory is FHS and is a
- constant -- The dep cache is named specifically now as 'PORTAGE_DEPCACHEDIR'.
-
- **Changed the stacking functions** so they are quite a bit more sane -- They
- still need a little help though -- stack_* functions are now used to stack
- specific types instead of using grab_stacked "super functions" -- The naming
- is a little rough but intelligable.
-
- **VIRTUALS modification** The /var/cache/edb/virtuals file is unnecessary
- as portage now loads the provides from the vartree itself. /etc/portage
- may have a virtuals file that stacks on top of the var and profile virtuals.
- This also entails the removal of the virtual-file handling code (yay!).
-
- 25 Apr 2004; Nicholas Jones <carpaski@gentoo.org> emerge: Fix for the name
- change for PORTAGE_CACHEDIR to PORTAGE_DEPCACHEDIR.
-
- 25 Apr 2004; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Added checks
- to ensure that use flags are properly mentioned in IUSE. Adjusted the QA
- Interceptors (function overrides for common app names) to make sure they
- don't interfere in weird cases -- Relying on 'type -p' now. Eclasses are
- checked for illegal inheritance modes (conditional-based). Increased the
- cache lines for an entry -- Added 'PROVIDE' and 8 empties.
-
- 22 Apr 2004; Marius Mauch <genone@gentoo.org> bin/repoman:
- added a CVS Header check to repoman
-
- 20 Apr 2004; Marius Mauch <genone@gentoo.org> pym/portage.py:
- fix for getmaskingreason if the mask isn't in PORTDIR's package.mask (#48447)
-
- 17 Apr 2004; Masatomo Nakano <nakano@gentoo.org> repoman: Fixed for
- python-2.2 compatibility.
-
-*portage-2.0.51_pre3 (13 Apr 2003): Cleanups and small features.
-
- 13 Apr 2004; Nicholas Jones <carpaski@gentoo.org> portage_db_*: Fixed the
- permissions issues regarding the umask problems with DB vars.
-
- 13 Apr 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Fix for the
- import urllib call that was misspelled for PORTAGE_BINHOST. Genone's patch
- that allows FETCHCOMMAND_${PROTOCOL}. getmaskingreadon() from genone's
- package.mask display patch.
-
- 13 Apr 2004; Nicholas Jones <carpaski@gentoo.org> output.py: For some reason
- people have a problem with my bad spelling of semi-common colors, so I added
- in requested changes for the color 'fuchsia'.
-
- 13 Apr 2004; Nicholas Jones <carpaski@gentoo.org> cnf/make.conf: Comment
- fixes for the CHOST line. FEATURES modification.
-
- 13 Apr 2004; Nicholas Jones <carpaski@gentoo.org> dispatch-conf.conf: Removed
- the 'a' option.
-
- 13 Apr 2004; Nicholas Jones <carpaski@gentoo.org> emerge: Catagory searches
- added if search is prefixed with '@'. jstubbs/spider's fix for binary use
- flag passing problems worked in. Genone's patch for package.mask comment
- display. Patch to display warnings when unmerging system packages worked in.
-
- 13 Apr 2004; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Quieted up
- some output for use/has internally. Added a patch for the help output.
- Added functionality to pass down IUSE from eclasses. Made several variables
- readonly inside of ebuilds:
- P PN PV PVR PR A AA D EBUILD EMERGE_FROM O PPID FILESDIR EBUILD_PHASE
-
-*portage-2.0.51_pre2 (11 Apr 2003): Release Fixes
-
- 11 Apr 2004; Nicholas Jones <carpaski@gentoo.org> emerge: Portage can now
- match categories if the search is prefixed with an '@' -- @app-portage will
- list all packages in app-portage... '@portage' will match all in app-portage
- and will match anything with portage in the title -- It's still a regex.
-
- 11 Apr 2004; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: nostrip in
- RESTRICT now enables DEBUGBUILD -- stripping and related functionality
- needs to get cleaned up better.
-
- 11 Apr 2004; Masatomo Nakano <nakano@gentoo.org> etc-update, ebuild.sh:
- Fixed infinity loop in etc-update(#19144). Fixed glob problem in
- ebuild.sh(#37066). Fixed deleting build-info problem with
- FEATURES="keepwork"(#29044).
-
- 10 Apr 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Includes the
- deadlock breaks marked with 'XXX:' to indicate where fixes are needed. Added
- HTTP encoding to PORTAGE_BINHOST with checks and fallbacks. Fallback for
- dbkey settings spawned from ebuild. Exception handling for db classes in
- the case of random corruption. jstubb's fix for the dep code to handle
- empty lists, added a notice about that being rude.
-
- 10 Apr 2004; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Made more QA
- notices and made them slightly more pretty 'QA Notice:'. Made the depend
- phase trap less violent -TERM not -KILL.
-
- 10 Apr 2004; Masatomo Nakano <nakano@gentoo.org> repoman: Added IUSE missings
- a check(#21544).
-
- 10 Apr 2004; Marius Mauch <genone@gentoo.org> tarball.sh, pym/portage.py:
- Fixing broken regexp in fixdbentries() (bug 46096), changing version
- number.
-
- 10 Apr 2004; Masatomo Nakano <nakano@gentoo.org> repoman: Fixed unsafety
- temporary file name(#44455).
-
- 09 Apr 2004; Masatomo Nakano <nakano@gentoo.org> portage.py: Fixed
- reorder ld.so.conf bug(#44028). Portage should skip not existing DIR
- during unmerge(#25339).
-
- 09 Apr 2004; Masatomo Nakano <nakano@gentoo.org> repoman: Added
- jstubbs's patch to add a check invalid DEPEND format to repoman.
- This should fix #36857.
-
- 09 Apr 2004; Masatomo Nakano <nakano@gentoo.org> emerge, portage.py,
- output.py: Fixed some bugs. See #45164, #24299, #34967.
-
- 01 Apr 2004; Masatomo Nakano <nakano@gentoo.org> portage.py: Fixed
- virtuals in reverse order(jstubbs's patch and my fix).
- This should close #45468.
-
- 20 Mar 2004; Nicholas Jones <carpaski@gentoo.org> portage_db_*: Updated
- all the db modules to be current and correct and updated the testing code
- to make sure everything is working properly.
-
- 20 Mar 2004; Nicholas Jones <carpaski@gentoo.org> ebuild: Fix for traceback
- and incorrent ROOT variable when using an alternate root and hand-merging.
-
- 20 Mar 2004; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Exported the
- sandbox variables so that they work.
-
- 20 Mar 2004; Nicholas Jones <carpaski@gentoo.org> emerge: Enhancements for
- binary downloads -- proper selection for virtuals, etc... Proper slot
- handling for pretend output with binaries.
-
- 20 Mar 2004; Nicholas Jones <carpaski@gentoo.org> prepman: symlink fix.
-
- 20 Mar 2004; Nicholas Jones <carpaski@gentoo.org> prepstrip: Added the
- readelf PIC code for the TEXTREL stuff.
-
- 20 Mar 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Removed some
- functions: evaluate, dep_listcleanup, dep_getjiggy. Made inheritance for
- profiles relative from the profile's directory instead of from the profile's
- base directory. OBSOLETING two functions dep_parenreduce and dep_opreduce in
- favor of the new code in portage_dep paren_reduce and use_reduce. dep_zapdeps
- got an overhaul for much more intelligent selection of packages and virtual
- handling during depgraph generation. Binary tree enhancements and selection
- enhancements. bindbapi created to do aux_get on binaries. isInjected added
- to vardbapi to get injected status. SRC_URI uses the new portage_dep code
- to handle strings now -- nesting should would flawlessly and FEATURES=cvs
- should get _all_ files.
-
- 20 Mar 2004; Nicholas Jones <carpaski@gentoo.org> portage_db_cpickle.py:
- Added pickle loading safety.
-
- 20 Mar 2004; Nicholas Jones <carpaski@gentoo.org> portage_dep: Temp home
- for the dep resolution code that will be moving from portage.py and emerge
- until we get the modular structure set up.
-
- 17 Mar 2004; Masatomo Nakano <nakano@gentoo.org> ebuild.sh, emerge,
- portage.py, portage_db_anydbm.py, portage_db_cpickle.py,
- portage_db_flat.py, portage_db_template.py:
- Improved handling cache files on multi portage trees.
- Fixed 'BAD COUNTER' error when emerge --inject. (#41062)
- Added ferringb's patch to avoid sed command. (#40819)
-
- 12 Mar 2004; Marius Mauch <genone@gentoo.org> bin/repoman:
- repoman: Added a readonly-variable-assignment check (#44424)
-
- 06 Mar 2004; Masatomo Nakano <nakano@gentoo.org> repoman, portage.py:
- Fixed some repoman/portage bugs. repoman shouldn't use /etc/portage/*
- files. repoman shouldn't use PORTDIR_OVERLAY(#11335). repoman should use
- each arch profile dir(#43601). portage didn't handle virtual dependency
- with version (>=virtual/package-1.0) in some places.
-
- 05 Mar 2004; Marius Mauch <genone@gentoo.org> bin/emerge:
- Trivial fix for emerge -pv if the download size is a long.
-
- 03 Mar 2004; Marius Mauch <genone@gentoo.org> bin/repoman,
- pym/portage.py:
- Fixing the "letter before endversion" bug (#17172). Replacing
- keywords.desc with arch.list in repoman (#35398). FEATURES=strict
- is now sufficient for Manifest validation (#41292).
-
- 01 Mar 2004; Marius Mauch <genone@gentoo.org> bin/emerge,
- bin/emergehelp.py, man/emerge.1:
- Updated docs for --update and removed the "help" action (wasn't working
- anyway).
-
- 29 Feb 2004; Masatomo Nakano <nakano@gentoo.org> portage.py: Fixed
- wrong USE in 'emerge info'. This should fix #34260.
-
- 27 Feb 2004; Masatomo Nakano <nakano@gentoo.org> emerge: Fixed info dir
- regeneration produces errors with a non-C locale and misleading error
- message. This should fix #41872,#24299.
-
- 27 Feb 2004; Masatomo Nakano <nakano@gentoo.org> emerge, emergehelp.py,
- emerge.1: Added genone's patch for man/help of --tree option.
-
- 27 Feb 2004; Masatomo Nakano <nakano@gentoo.org> ebuild.sh: Removed
- /usr/share directory in ${D} when it's empty. This should close #42312.
-
- 26 Feb 2004; Masatomo Nakano <nakano@gentoo.org> emerge: Added new option
- --newuse. This option is to rebuild a package whose USE has been changed.
-
- 22 Feb 2004; Masatomo Nakano <nakano@gentoo.org> emerge, portage.py: speedup
- when /etc/portage/package.keywords is defined. Moved loading
- /etc/portage/package.* processs to config class. This should fix #41520.
-
- 19 Feb 2004; Masatomo Nakano <nakano@gentoo.org> portage.py:
- Fixed CATEGORY value after preinst phase. This should close #6414. Fixed
- nested dependency problem and cleaned up dep_zapdeps function.
- This bug happened with DEPEND='|| ( cat_a/pkg_a flag? ( cat_b/pkg_b ) )'.
- This should close #41869.
-
- 13 Feb 2004; Masatomo Nakano <nakano@gentoo.org> portage.py: Fixed parsing
- SRC_URI bug when FEATURES="cvs". This should close #16159.
-
- 12 Feb 2004; Masatomo Nakano <nakano@gentoo.org> emerge, pym/portage.py:
- TGL's patch for correction package size when emerge -v. -- Fixed
- use.default bug. It occurs when package in use.default exists in system
- and it's virtual package. This should close #40831.
-
- 12 Feb 2004; Masatomo Nakano <nakano@gentoo.org> emerge: Fixed
- --ask bugs. It breaks with "--clean". It also breaks when blocker
- exists. This should close #39865.
-
- 12 Feb 2004; Masatomo Nakano <nakano@gentoo.org> emerge: Fixed
- --ask bug with --changelog. This should close #41293.
-
- 11 Feb 2004; Masatomo Nakano <nakano@gentoo.org> emergehelp.py: Added
- help of F flag with emerge --pretend. This should close #28253.
-
- 11 Feb 2004; Masatomo Nakano <nakano@gentoo.org> pym/portage.py: Fixed
- ccache dir permission problem with FEATURES="userpriv".
- This should fix #22125.
-
-*portage-2.0.50-r1 (09 Feb 2003): Release Fixes
-
- 09 Jan 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Fix the config
- code so that it doesn't fail when the profile does not exist -- allows
- sync'ing without a tree like it should. Parser returns an exception with
- the parse error now for getconfig(). TGL's patch for another cache issue
- in class config. Modified the /etc/make.profile message. Fix for the
- "eclass does not exist" messages on sync.
-
- 09 Jan 2004; Nicholas Jones <carpaski@gentoo.org> repoman: Fix the repopath
- so that it cna be run outside of cvs repos.
-
- 09 Jan 2004; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Add portage's
- pym path to SANDBOX_PREDICT to stunt any further .pyc/.pyo problems. Add
- nofetch to the 'successful' kill list to stop the $T definition woes.
-
- 08 Feb 2004; Masatomo Nakano <nakano@gentoo.org> repoman: repoman should
- read each arch virtual file. This should close #40813.
-
- 08 Feb 2004; Masatomo Nakano <nakano@gentoo.org> emerge: Fixed -s/-S
- bug. "Latest version installed:" was incorrect. This should fix #40847
-
- 08 Feb 2004; Masatomo Nakano <nakano@gentoo.org> portage.py: Fixed ldconfig
- bug. emerge didn't ldconfig after removing library directory.
- This should fix #40694.
-
-*portage-2.0.50 (06 Feb 2003): Release -- API change, cleanups, speedups
-
- 06 Jan 2004; Nicholas Jones <carpaski@gentoo.org> *: repoman got a quick fix
- from genone. Ed's fix for ask/pretend. Made sure that emerge force-updated
- the eclass cache before trying to update all the metadata. masking type
- patch from Genone. masking info patch and regenworld patch added.
-
-*portage-2.0.50_pre22 (04 Feb 2003): Cleanups and stablizing
-
- 04 Feb 2004; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: TGL's fixes
- for exec/child/wait problems. Unset GREP_OPTIONS GREP_COLOR. has() and use()
- no longer attempt to determine if they are to be quiet or noisy -- They
- default to noisy -- useq() and hasq() are the non-verbose versions.
- EBUILD_PHASE set to add a hack-ish way around global scope calls in
- eclasses -- NOTHING SHOULD BE CALLED IN THE GLOBAL SCOPE. Touchup to the
- inherit() code that should finally allow the removal of the ECLASS and
- INHERITED settings. Removed tty (use/has) calls. Removed dirname calls --
- portage.py handles setting the dbkey filename now.
-
- 04 Feb 2004; Nicholas Jones <carpaski@gentoo.org> emerge: Ed Catmur's
- (with a little TGL added in) patch for --ask. Added a 'metadata' target
- that skips the sync and only updates the cache. FEATURES="getbinpkg" added.
- TGL's exit code fixes. Fixed match code for -S so it doesn't complain about
- specific and double versions. Unmerge via dbpath fix. Rewrote rsync's
- options that supports --verbose and --quiet operation now and can force
- checksumming all files using --debug. Sort the files in the cache update
- so it's a little more predictable.
-
- 04 Feb 2004; Nicholas Jones <carpaski@gentoo.org> prepstrip: changed
- --strip-debug to --strip-unneeded.
-
- 04 Feb 2004; Nicholas Jones <carpaski@gentoo.org> getbinpkg.py: Updates to
- enable HTTP/HTTPS authentication.
-
- 04 Feb 04; Nicholas Jones <carpaski@gentoo.org> portage.py: best_from_dict
- added to grab the best entry from set of dicts using a list of the keys for
- priority. jstubb's patch to fix listdir -- splits it into a cache and list
- setup. jstubb's patch for varexpand to handle $VAR better. Latexer's patch
- for KernelVersion code to use Makefiles instead of the version.h. Modules
- are loaded from /etc/portage/modules or defaults, whichever works. Fixed
- the /etc/make.profile-is-missing traceback. Spawn can be given 3 pipes to
- redirect stdin,stdout,stderr to specific outputs, terminals, or files.
- TGL's patch for cache functions in portage.py so that they do not cache at
- inappropriate times. PORTAGE_TMPFS is now used if set as a temporary file
- operation area -- recommended to actually be a ramfs/tmpfs filesystem for
- speed. Genone enhanced the deprecated profile patch.
-
- 31 Jan 2004; Masatomo Nakano <nakano@gentoo.org> emerge: Fixed --skipfirst
- bug. This closes #36880.
-
- 29 Jan 2004; Masatomo Nakano <nakano@gentoo.org> emerge: TGL's patch
- for imporving overlay verbose. This closes #39765.
-
- 27 Jan 2004; Masatomo Nakano <nakano@gentoo.org> portage.py: Fixed
- autouse bug. autouse were ignored.
-
- 21 Jan 2004; Nicholas Jones <carpaski@gentoo.org> emerge: Output failed
- cache updates during emerge sync.
-
- 21 Jan 2004; Nicholas Jones <carpaski@gentoo.org> *: VDB_PATH fixes.
-
- 21 Jan 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Only use
- custom profiles when not called by repoman. ROOT never changes profile
- roots, only custom/system profiles var/cache/edb/virtuals. Sandbox fix
- where sandbox was creating an invalid logfile (not giving a summary)
- due to a '/' in SANDBOX_LOG. Turned down the Lockfile output. Double
- check the INCOMPLETE MERGE identifications as it can be caused by cache.
-
- 24 Jan 2004; <nakano@gentoo.org> emerge: Improved timestamp check
- when 'emerge sync'. Added catching amiguous error when unmerge.
- This closes #24325.
-
- 23 Jan 2004; <nakano@gentoo.org> emerge, portage.py: Fixed 2 bugs.
- Portage doesn't read local virtuals file, which happens on only cvs
- version. package is blocked by itself.
-
- 21 Jan 2004; Nicholas Jones <carpaski@gentoo.org> portage.py, emerge:
- Fix from genone for emerge's direct reading of packages and his patch
- that also adds in /etc/portage/profile as a stacked profile.
-
- 21 Jan 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Completed
- inheritence capabilities for portage.config reading some files. Moved a
- copy of the getvirtuals() function into settings to handle multiple
- profiles properly.
-
-*portage-2.0.50_pre17/18/19 (21 Jan 2004): Modules for DBs and quick fixes
-
- 21 Jan 2004; Nicholas Jones <carpaski@gentoo.org> *: Moved all references
- to var/db/pkg to portage.VDB_PATH --- This will change again -- NEED TO
- BE MOVED INTO A PATH/CONSTANTS SETUP.
-
- 21 Jan 2004; Nicholas Jones <carpaski@gentoo.org> dosed: Quick fix for
- the basename missing/misplaced issue.
-
- 21 Jan 2004; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Added
- /dev/console to PREDICT to attempt a workaround for a serial console
- bug. dbkey is now set through portage.py/doebuild to allow for modular
- db code.
-
- 21 Jan 2004; Nicholas Jones <carpaski@gentoo.org> emerge: regen doesn't
- require root anymore. Edited the timestamp check to be a little more
- friendly -- delete the portdir timestamp and it won't use the alternate.
- Fix some permission settings. Added some warnings in for cachedirs that
- are very likely to ruin your system. Cleaned out some of the eclass code
- that isnt valid any longer.
-
- 21 Jan 2004; Nicholas Jones <carpaski@gentoo.org> portageq: Added vdb_path
- as a target to get the db directory. Quickpkg uses this.
-
- 21 Jan 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Added
- load_mod() -- grabs a class/function from a module and passes it back
- without loading the module into the global scope. Added unique_array()
- which eliminates duplicates from an array. grab_stacked() operates like
- the other grab* and getconfig functions, but takes a filename and a set
- of paths that it will apply incrementally or clobbers -- for profile
- inheritance. getconfig no longer exits on non-existance returns None.
- Class config now should be passed a profile path and a set of incremental
- values instead of using the globals -- defaults to using the globals
- presently and print an error message. Adding support for module configs
- as a set of strings 'class.subclass.objectmodule':'module.to.use.object'
- for load_mod and the database modules. Profile inheritance started. Killed
- the eclass() super-function and replaced it with class eclass_cache that
- is visible and conceptually simpler -- Also uses the plugable modules.
- Cleaned out the sync calls for the DBs. MASSIVE simplification of the
- aux_get code -- removed memory-caching in favor of system cache (actually
- faster in all cases so far -- P100 and P4-2.2G). Lockfile usage around the
- cachefile.
-
- 21 Jan 2004; Nicholas Jones <carpaski@gentoo.org> portage_db_*: Updated
- the API a little but to have permissions set properly. A little more
- reorganization and removed the keycount checks.
-
- 21 Jan 2004; Masatomo Nakano <nakano@gentoo.org> emerge: download size
- should not be displayed when the package is nomerge with --tree.
-
- 20 Jan 2004; Masatomo Nakano <nakano@gentoo.org> portage.py: Reverted
- ambiguity package fix in cpv_expand().
-
- 20 Jan 2004; Masatomo Nakano <nakano@gentoo.org> emerge: Moved
- backup timestamp.chk file from portage tree to PORTAGE_TMPDIR.
-
- 20 Jan 2004; Masatomo Nakano <nakano@gentoo.org> emerge: Reverted the
- backing up the timestamp.chk fix.
-
- 20 Jan 2004; Masatomo Nakano <nakano@gentoo.org> emerge: Improved the
- list of --tree by TGL's patch. This should close #38070.
-
- 20 Jan 2004; Masatomo Nakano <nakano@gentoo.org> emerge: Removed debug
- message without --debug. This should close #23840.
-
- 19 Jan 2004; Masatomo Nakano <nakano@gentoo.org> emerge: Improved
- timestamp check of rsync. This should close #37403.
-
- 19 Jan 2004; Masatomo Nakano <nakano@gentoo.org> portage.py: Improved
- regeneration ld.so.cache. This should close #37858.
-
- 19 Jan 2004; Masatomo Nakano <nakano@gentoo.org> emerge: Fixed bug which
- emerge doesn't block same package but different version.
- (example: DEPEND="!<cat/pkg-1.0.0" in cat/pkg-1.0.0.ebuild)
-
- 19 Jan 2004; Masatomo Nakano <nakano@gentoo.org> portage.py:
- Modified cpv_expand() to check package.mask. This should close #38592.
-
- 19 Jan 2004; Masatomo Nakano <nakano@gentoo.org> portage.py: Escaped
- regualar expression for replace entry in fixdbentries().
-
- 18 Jan 2004; Masatomo Nakano <nakano@gentoo.org> portage.py:
- Fixed AUTOCLEAN delay problem in .50pre* by TGL's patch. This close
- #38189. Fixed unmerge failture bug when 'ebuild foo-1.0.0 unmerge'.
- These close #38189, #38366
-
- 18 Jan 2004; Masatomo Nakano <nakano@gentoo.org> emerge, portage.py:
- Fixed "ebuild /foo/bar-1.0.0.ebuild unmerge" and "emerge bar-1.0.0 unmerge"
- problems. This should close #38420.
-
- 17 Jan 2004; Masatomo Nakano <nakano@gentoo.org> portage.py: Fixed
- "!<=" style block problem. Fixed symlink with absolute path
- problem in treewalk().
-
-*portage-2.0.50_pre16 (13 Jan 2004): Quick Fixes -- ~arch version
-
- 13 Jan 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Removed an
- unnecessary depend call that double eclass-using ebuild's cache regen
- time.
-
-*portage-2.0.50_pre15 (12 Jan 2004): Quick Fixes -- ~arch version
-
- 12 Jan 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Unmerge
- traceback fix.
-
-*portage-2.0.50_pre14 (12 Jan 2004): Quick Fixes -- ~arch version
-
- 12 Jan 2004; Nicholas Jones <carpaski@gentoo.org> emerge: Fix for
- traceback on '-S'.
-
- 12 Jan 2004; Nicholas Jones <carpaski@gentoo.org> repoman: Fix for
- traceback on --help.
-
- 12 Jan 2004; Nicholas Jones <carpaski@gentoo.org> sandbox: Fix for
- sandboxpids.tmp file accesses.
-
- 12 Jan 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Sandbox, as
- above. Catch invalid package names and print a sane message about it.
-
-*portage-2.0.50_pre13 (11 Jan 2004): Fixes
-
- 11 Jan 2004; Nicholas Jones <carpaski@gentoo.org> cnf/*: Updated the
- Advanced masking section to aid the reduction of user complaints and
- requests for unreasable usage of ACCEPT_KEYWORDS.
-
- 11 Jan 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: A counter
- fix was fixed to actually check the counters of all CP versions to ensure
- the new counter is higher than all existing ones. Modified the dblink
- class to have class lockfiles for the db and tmpdb dirs as well as lock
- other files before editing. Reorganization of the merge code in dblink
- so that the tmpdb is filled immediately after preinst and prior to the
- actual FS merging -- COUNTER and CONTENTS go directly into the tmpdb
- and not into the infodir.
-
-*portage-2.0.50_pre11/12 (09 Dec 2003): repoman/binpkg/exit conditions
-
- 09 Jan 2004; Nicholas Jones <carpaski@gentoo.org> emerge: getbinpkgonly
- fixes for emerge -G world, should behave properly now instead of using
- ebuild masks. Only downloads immediately before a merge -- fetchonly now
- applies to binary packages.
-
- 08 Jan 2004; Masatomo Nakano <nakano@gentoo.org> repoman: Ignore other
- arches check in repoman when --ignore-other-arches(-I).
-
-*portage-2.0.50_pre10 (06 Dec 2003): API change + enhancements
-
- 06 Jan 2004; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Fix for
- dyn_preinst being called before IMAGE was set -- IMAGE is now valid
- in pkg_preinst. Added suidctl for SELinux.
-
- 06 Jan 2004; Nicholas Jones <carpaski@gentoo.org> emerge: Added -P to
- initial cvs checkout.
-
- 06 Jan 2004; Nicholas Jones <carpaski@gentoo.org> quickpkg: Fix for
- the 'tar up /' problem.
-
- 06 Jan 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Caught a
- traceback generated by bad depend atoms for repoman. Fixes from genone
- for package.*. Fixed the checks for doebuild calls in treewalk that was
- ignoring exit conditions for ebuilds.
-
- 04 Jan 2004; Masatomo Nakano <nakano@gentoo.org> repoman: Added PDEPEND
- dependency check. This closes #24796
-
- 04 Jan 2004; Masatomo Nakano <nakano@gentoo.org> repoman, portage.py:
- Added new dependency check to repoman. This closes #36887.
-
- 03 Jan 2004; Masatomo Nakano <nakano@gentoo.org> emerge: Modified
- to specific port number in emerge sync. This closes #36994
-
- 02 Jan 2004; Masatomo Nakano <nakano@gentoo.org> portage.py: Fixed
- a problem that emerge doesn't block package when it's required.
- It happens in .50_pre*.
-
- 02 Jan 2004; Masatomo Nakano <nakano@gentoo.org> portage.py: Fixed
- issue with getsize() when --debug.
-
- 02 Jan 2004; Masatomo Nakano <nakano@gentoo.org> portage.py: Fixed
- issue with virtual. This closes bug #9050, #22225, #29499.
-
- 01 Jan 2004; Masatomo Nakano <nakano@gentoo.org> ebuild, emerge, portage.py:
- Fixed issue with not cleaning up temp directory. This closes bug #34967.
-
- 31 Dec 2003; Masatomo Nakano <nakano@gentoo.org> emerge:
- Fixed 'emerge sync' issue which continuously connects to same host.
-
- 31 Dec 2003; Nicholas Jones <carpaski@gentoo.org> emerge: Found the line
- that was causing the package dir to be printed... It was a spawn call.
-
- 31 Dec 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Fix for
- the symlink corruption in the db from the movefile() bug.
-
- 29 Dec 2003; Masatomo Nakano <nakano@gentoo.org> portage.py:
- Fixed bug which emerge stops when no denpendencies exist in || ( )
- by USE flags. This closes #36568.
-
- 29 Dec 2003; Masatomo Nakano <nakano@gentoo.org> emerge, portage.py:
- Added an ambiguity package check when emerge. This closes bug #22700.
-
-*portage-2.0.50_pre9 (24 Dec 2003): API change + enhancements
-
- 24 Dec 2003; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Added
- PORTAGE_TMPDIR to SANDBOX_READ/WRITE to ensure it works. SpanKY's
- patch for use negation added (use !foo). pkg_setup doesn't die on
- a non-zero exit status.
-
- 24 Dec 2003; Nicholas Jones <carpaski@gentoo.org> emerge: using os.uname
- instead of calling out to uname.
-
- 24 Dec 2003; Nicholas Jones <carpaski@gentoo.org> quickpkg: Added SpanKY's
- patch for delayed exit/error conditions.
-
- 24 Dec 2003; Nicholas Jones <carpaski@gentoo.org> xpak.py: chdir's added
- to the getcwd fix for missing dirs.
-
- 24 Dec 2003; Masatomo Nakano <nakano@gentoo.org> emerge: Added OVERLAY
- directories display for --verbose.
-
-*portage-2.0.50_pre8 (24 Dec 2003): API change + enhancements
-
- 22 Dec 2003; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Added
- /proc/self/maps to SANDBOX_PREDICT, and /dev/shm to read/write.
-
- 22 Dec 2003; Nicholas Jones <carpaski@gentoo.org> emerge: Added automake
- and autoconf versions to the output of emerge info.
-
- 22 Dec 2003; Nicholas Jones <carpaski@gentoo.org> etc-update: Added
- edit merged file option -- defaults to EDITOR var or "nano -w".
-
- 22 Dec 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Use
- os.uname instead of calling out to uname which might not exist.
-
-*portage-2.0.50_pre7 (22 Dec 2003): API change + enhancements
-
- 22 Dec 2003; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: patch to
- quote most of the path operators that might involve spaces.
-
- 22 Dec 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Fix for
- invalid entries in package.keywords. Character chopping on mirrors
- fixed again.
-
- 21 Dec 2003; Masatomo Nakano <nakano@gentoo.org> bin/ebuild, bin/emerge,
- pym/portage.py: Changed to show disabled USE flags from use.mask when
- using emerge -vp. And fixed use.mask issue.
-
- 20 Dec 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Rewrote
- match_from_list -- Simplified and made pkgcmp and match_from_list
- properly compare package names.
-
- 20 Dec 2003; Nicholas Jones <carpaski@gentoo.org> repoman: Fix for mysigs
- traceback when signing.
-
- 20 Dec 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Added
- PYTHONPATH to the specials list -- created a colon_seperated list.
- Fixed reset() in class config so that you can specify keeping the
- pkg dictionary when resetting the values.
-
- 19 Dec 2003; Masatomo Nakano <nakano@gentoo.org> repoman: Added check
- whether "ebuild foo.ebuild digest" succeeds.
-
- 19 Dec 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Fix for
- pkg settings being maintained after an unmerge.
-
- 19 Dec 2003; Nicholas Jones <carpaski@gentoo.org> pym/portage_db_*: Moved
- to using cPickle instead of marshal. More standardization of the API.
-
- 18 Dec 2003; Masatomo Nakano <nakano@gentoo.org> repoman: Added virtual
- dependency check on each arch.
-
- 17 Dec 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Fixed a
- permission issue involving $T and userpriv. Lockfile touchup.
-
- 17 Dec 2003; Nicholas Jones <carpaski@gentoo.org> portage_db_*: Added
- templates and db for cache interfaces. Presently have a anydbm and a
- flat file interface working. See the test for operations.
-
- 15 Dec 2003; Nicholas Jones <carpaski@gentoo.org> emerge: Added a call
- to portageq that causes python to create optimized modules prior to it
- ending up inside the sandbox. Added more output and logging to sync.
-
- 15 Dec 2003; Nicholas Jones <carpaski@gentoo.org> prepstrip: 'tree' is not
- the same as 'true'.
-
- 15 Dec 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: invalid
- settings in package.keywords caused a traceback -- fixed with error message.
-
-*portage-2.0.50_pre1 (12 Dec 2003): API change + enhancements
-
- 10 Dec 2003; Nicholas Jones <carpaski@gentoo.org> chkcontents: Uses portage
- functions to do md5sum calcs.
-
- 10 Dec 2003; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Removed try()
- as it isn't used, and was deprecated for a long while. Genone's fetching
- size display added for --verbose. License display added. Added a little
- debug for IUSE so we can figure out the binary package --verbose IUSE
- issues that are randomly reported. XXXXXXXXXXXXXXXXXXX's 'buildsyspkg'
- patch for building only system packages into tbz2s. Unmerge fix for new
- settings instances. RSYNC_RATELIMIT added.
-
- 10 Dec 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: ADA path
- variables added to specials for env_update. Error messaeg correction for
- make.defaults syntax errors. Unmerge now uses the environment file, if it
- exists, to get the complete environment back to perform unmerge operations.
- load_infodir() uses pkg settings completely now. Fixed the passing of
- settings in unmerge and dblink. Fixed an issue regarding unlinking lockfiles
- while inside of a sandbox.
-
- 09 Dec 2003; Nicholas Jones <carpaski@gentoo.org> ebuild.sh, *.sh:
- Moved helper scripts into bin/functions and made them sourceable -- they
- now will die in cases where sub-parts fail. dodoc and keepdir are now
- recursive-capable.
-
- 09 Dec 2003; Nicholas Jones <carpaski@gentoo.org> emerge: emerge.log now
- set as portage:portage with 0660 perms. --debug now enables tracebacks
- for dep generation instead of moving code out of the try block.
-
- 09 Dec 2003; Nicholas Jones <carpaski@gentoo.org> g-cpan.pl: rac's patch
- to get arch list from portage's list of arches in the profiles.
-
- 09 Dec 2003; Nicholas Jones <carpaski@gentoo.org> repoman: Moved a bit of
- the existing gpg code around -- it might work as is, but requires 'sign'
- in features. Fixed a potential for repoman to miss updates that should
- get a new manifest and commit. Fixed digest/manifest generation for
- non-packagedir runs of repoman.
-
- 09 Dec 2003; Nicholas Jones <carpaski@gentoo.org> emergehelp.py, make.conf,
- getbinpkg.py: Message touch ups.
-
- 09 Dec 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: load_infodir()
- uses pkg settings now instead of env and backup. Genone's custom mirror
- patch included. Added some missing 'strict' flags for recursion in digest*().
- Refixed the invalidentry stuff that was lost across patch merges. Fix for
- pkg-keywords from genone included. Genone's deprecated profile patch for
- reporting to a user that their current profile is deprecated. Message about
- missing arch.list instead of spouting invalid keywords messages.
-
- 08 Dec 2003; Masatomo Nakano <nakano@gentoo.org> repoman:
- Added all arch dependency check. This closes bug #24160.
-
- 07 Dec 2003; Masatomo Nakano <nakano@gentoo.org> emerge,portage.py:
- Fixed bugs. 1.--debug doesn't work 2.Portage breaks files
- in /var/db/*/*. 3.No stop if dependency problem happens.
- They are only cvs version problems.
-
- 01 Dec 2003; Masatomo Nakano <nakano@gentoo.org> emerge: Fixed bug which
- always remakes info dir file.
-
- 29 Nov 2003; Masatomo Nakano <nakano@gentoo.org> portage.py: Fixed issue with
- ebuild name rule. Fixed typo with variable name.
- This closes bug #17172,#34666
-
- 29 Nov 2003; Masatomo Nakano <nakano@gentoo.org> emerge: Fixed issue with
- lacking the "setting" argument for pkgmerge()
-
- 29 Nov 2003; Masatomo Nakano <nakano@gentoo.org> emerge: fixed rsync bug.
- This closes bug #34660.
-
- 28 Nov 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Migration
- to non-global settings started -- class config has new functionality and
- is locked after portage is finished initializing -- changes cannot be made
- to the global instance of config 'settings' -- reset() is now functional,
- setcpv() loads PKGUSE from /etc/portage/package.use, load_infodir() loads
- all small files (under 4k) from the vardb directory of an installed package
- so that operations have the same post* settings as they had at merge time.
- Begin modifications to spawn() to allow for files/pipes to be used for
- IO instead of using getstatusoutput which does not take an environment
- parameter like execve(). check_config_instance() ensures that the provided
- parameter is a 'class config' instance -- for ensuring that everything is
- being passed properly with the changes. Fix for the local FS mirror issue
- where it removed the first '/' instead of the last one. doebuild() cleanups
- for readability and pkguse enhancements -- also remove getstatusoutput()
- usage for depend so that we don't have to modify the active environment.
- Fix for symlink mtime values returned from movefile. (Nakano) SLOTMOVE
- added to global update functionality to fix some issues where a package
- suddenly must become slotted. portdbapi takes a root parameter instead
- of using settings. Slightly more useful output from depend. binarytree()
- now takes a pkgdir instead of using settings. Portage will now die if
- ebuild.sh exits on a signal.
-
- Moved some functions around and renamed them for general use -- derived
- from match2 in class portagetree:
- match_to_list() find all atoms in a list that match a given package.
- best_match_to_list() determines the most specific match. Needs work.
- match_from_list() find all packages in a list that match a given atom.
-
- 28 Nov 2003; Nicholas Jones <carpaski@gentoo.org> emerge: Fixed an issue
- with searchdesc wanting root permissions if run as non-root. Migrated to
- the non-global config class. EMERGE_FROM added for the dyn_preinst patch
- -- Indicates if a merge is occuring from an ebuild or from a binary. Patch
- for rsync timestamp checking from Nakano.
-
- 28 Nov 2003; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Save PKGUSE.
- Pebeneto's patch for dyn_preinst and SELinux added as a fix for binary
- and ebuild merges. Added a kill for portage during the depend phase so
- that portage will actually die if you control-C.
-
- 28 Nov 2003; Nicholas Jones <carpaski@gentoo.org> *: MASSIVE set of changes
- to start using locally defined 'class config' instances. This allows us to
- start working on some parallelism among other things. Created this way:
- mysettings = portage.config(clone=portage.settings)
-
- The Following functions now take a 'config' parameter:
- spawn(), fetch(), digestgen(), digestcheck(), spawnebuild(), doebuild(),
- merge(), dep_opconvert(), dep_check(), dblink.__init__()
-
- package.keywords is now implemented curtasy of genone/max. PKGUSE was
- rewritten for the global config killing and is also included. X11 man
- pages now found and zipped correctly. SYS.PATH fixes for the python
- migration -- issue actually only shows up on 2.2 systems because of how
- compiled modules are used if found regardless of the original source's
- existance.
-
- 28 Nov 2003; Nicholas Jones <carpaski@gentoo.org> tabcheck.py: An easier
- way to make sure that all the python stuff is correctly using tabs and
- not mixing spaces.
-
- 28 Nov 2003; Nicholas Jones <carpaski@gentoo.org> xpak, xpak.py: Fixes
- to ensure that it works if the current dir is missing and that the python
- path gets set properly.
-
- 22 Nov 2003; Daniel Robbins <drobbins@gentoo.org> portage.py: Fixed
- calls in vartree method to invalidentry().... made them call call
- self.dbapi.invalidentry() (there were multiple wrong method calls.)
-
- 10 Nov 2003; Nicholas Jones <carpaski@gentoo.org> md5check.py: Checks all
- digests and SRC_URIs for filenames and associated MD5s. Reports collisions
- between versions/packages, missing, and extra lines in digests.
-
-*portage-2.0.49-r17/18 (10 Nov 2003): Fixes
-
- 10 Nov 2003; Nicholas Jones <carpaski@gentoo.org> *: Changed portage to
- be the first path in sys.path for all python scripts. Also enabled
- optimizations from the scripts to ensure everything imported is built
- for speed. ebuild: applied fix for the '//' root breaking the db[].
- prepstrip: etdyn quickfix
-
- 10 Nov 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: fix for
- RESTRICT=nouserpriv. GENTOO_MIRRORS can have paths set to take files
- from. Fixes for mishandled cache data regarding *pkgsplit(). Fixes for
- '*' being returned as part of a package split. An 'invalidentry()' fix
- for a traceback. Nakano's fixes for virtual removals not working properly,
- sandbox violations during pkg_nofetch, || depend selection. Genone's
- fixpackages speedup.
-
- 10 Nov 2003; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: PORTAGE_TMPDIR
- fix for distcc. Variable passing bug patch for export_functions. Exit 1
- added for nofetch to stop sandbox violation. Nakano's --tree patch added.
- Improved the unmerge messages to denote what kind of unmerge fails. Info
- pages regex pattern adjusted to allow most any name for a page.
-
- 10 Nov 2003; Nicholas Jones <carpaski@gentoo.org> repoman: genone's xml
- linting additions.
-
- 01 Nov 2003; Robin H. Johnson <robbat2@gentoo.org> pym/cvstree.py:
- fix bug #32071, by properly escaping a string to not be a regex. Checked
- thru entire *.py tree and found this is the only mis-use of strings that
- need to be escaped.
-
- 31 Oct 2003; Daniel Robbins <drobbins@gentoo.org> portage.py: /lib/modules
- now gets "unmerge protection." This is half of the config protection
- functionality. It means that anything in /lib/modules will not be deleted
- when a package is unmerged (often automatically when a user merges a
- kernel module ebuild for a new kernel.) This solves the "my module
- disappeared!" issue. This closes bug #1477.
-
- 31 Oct 2003; Daniel Robbins <drobbins@gentoo.org> emerge: Should no longer
- spit out wacky "!!! no match found" warnings when auto-cleaning.
-
- 30 Oct 2003; Daniel Robbins <drobbins@gentoo.org> portage.py: Only run
- depscan.sh if it exists on disk. This allows Portage to run inside a stage1
- where /sbin/depscan.sh doesn't exist.
-
- 30 Oct 2003; Daniel Robbins <drobbins@gentoo.org> portage.py: Applied fix to
- allow multi-level "use? ( )" in SRC_URI, closing bug #16159.
-
-*portage-2.0.49-r15/16 (21 Oct 2003): Fixes
-
- 21 Oct 2003; Nicholas Jones <carpaski@gentoo.org> fix-db.py: was broken
- for python2.3 -- fixed now.
-
- 21 Oct 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Added
- lockfiles to prelink md5 checks. Fixed caching bug where cache objects
- were passed back as pointers instead of copies. Added 'invalidentry'
- function to handle lockfiles -- It tests/deletes them using unlockfile.
- Added fix-db.py to the 'databases is broken' messages.
-
- 21 Oct 2003; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: added CDPATH
- to unset. SELinux fix for sandbox.
-
+++ /dev/null
-# ChangeLog for Portage; the Gentoo Linux ports system
-# Copyright 1999-2004 Gentoo Foundation; Distributed under the GPL v2
-# $Id: ChangeLog.000,v 1.3 2004/10/04 13:58:57 vapier Exp $
-
- *** THIS IS FOR ARCHIVAL PURPOSES ONLY -- DO NOT MODIFY ***
- *** THIS IS FOR ARCHIVAL PURPOSES ONLY -- DO NOT MODIFY ***
- *** THIS IS FOR ARCHIVAL PURPOSES ONLY -- DO NOT MODIFY ***
- *** THIS IS FOR ARCHIVAL PURPOSES ONLY -- DO NOT MODIFY ***
- *** THIS IS FOR ARCHIVAL PURPOSES ONLY -- DO NOT MODIFY ***
- *** THIS IS FOR ARCHIVAL PURPOSES ONLY -- DO NOT MODIFY ***
- *** THIS IS FOR ARCHIVAL PURPOSES ONLY -- DO NOT MODIFY ***
- *** THIS IS FOR ARCHIVAL PURPOSES ONLY -- DO NOT MODIFY ***
- *** THIS IS FOR ARCHIVAL PURPOSES ONLY -- DO NOT MODIFY ***
-
- 21 Jan 2004; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: TGL's fixes
- for exec/child/wait problems. Unset GREP_OPTIONS GREP_COLOR. has() and use()
- no longer attempt to determine if they are to be quiet or noisy -- They
- default to noisy -- useq() and hasq() are the non-verbose versions.
- EBUILD_PHASE set to add a hack-ish way around global scope calls in
- eclasses -- NOTHING SHOULD BE CALLED IN THE GLOBAL SCOPE. Touchup to the
- inherit() code that should finally allow the removal of the ECLASS and
- INHERITED settings. Removed tty (use/has) calls. Removed dirname calls --
- portage.py handles setting the dbkey filename now.
-
- 21 Jan 2004; Nicholas Jones <carpaski@gentoo.org> emerge: Ed Catmur's
- (with a little TGL added in) patch for --ask. Added a 'metadata' target
- that skips the sync and only updates the cache. FEATURES="getbinpkg" added.
- TGL's exit code fixes. Fixed match code for -S so it doesn't complain about
- specific and double versions. Unmerge via dbpath fix. Rewrote rsync's
- options that supports --verbose and --quiet operation now and can force
- checksumming all files using --debug. Sort the files in the cache update
- so it's a little more predictable.
-
- 21 Jan 2004; Nicholas Jones <carpaski@gentoo.org> prepstrip: changed
- --strip-debug to --strip-unneeded.
-
- 21 Jan 2004; Nicholas Jones <carpaski@gentoo.org> getbinpkg.py: Updates to
- enable HTTP/HTTPS authentication.
-
- 21 Jan 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: best_from_dict
- added to grab the best entry from set of dicts using a list of the keys for
- priority. jstubb's patch to fix listdir -- splits it into a cache and list
- setup. jstubb's patch for varexpand to handle $VAR better. Latexer's patch
- for KernelVersion code to use Makefiles instead of the version.h. Modules
- are loaded from /etc/portage/modules or defaults, whichever works. Fixed
- the /etc/make.profile-is-missing traceback. Spawn can be given 3 pipes to
- redirect stdin,stdout,stderr to specific outputs, terminals, or files.
- TGL's patch for cache functions in portage.py so that they do not cache at
- inappropriate times. PORTAGE_TMPFS is now used if set as a temporary file
- operation area -- recommended to actually be a ramfs/tmpfs filesystem for
- speed. Genone enhanced the deprecated profile patch.
-
- 31 Jan 2004; Masatomo Nakano <nakano@gentoo.org> emerge: Fixed --skipfirst
- bug. This closes #36880.
-
- 29 Jan 2004; Masatomo Nakano <nakano@gentoo.org> emerge: TGL's patch
- for imporving overlay verbose. This closes #39765.
-
- 27 Jan 2004; Masatomo Nakano <nakano@gentoo.org> portage.py: Fixed
- autouse bug. autouse were ignored.
-
- 21 Jan 2004; Nicholas Jones <carpaski@gentoo.org> emerge: Output failed
- cache updates during emerge sync.
-
- 21 Jan 2004; Nicholas Jones <carpaski@gentoo.org> *: VDB_PATH fixes.
-
- 21 Jan 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Only use
- custom profiles when not called by repoman. ROOT never changes profile
- roots, only custom/system profiles var/cache/edb/virtuals. Sandbox fix
- where sandbox was creating an invalid logfile (not giving a summary)
- due to a '/' in SANDBOX_LOG. Turned down the Lockfile output. Double
- check the INCOMPLETE MERGE identifications as it can be caused by cache.
-
- 24 Jan 2004; <nakano@gentoo.org> emerge: Improved timestamp check
- when 'emerge sync'. Added catching amiguous error when unmerge.
- This closes #24325.
-
- 23 Jan 2004; <nakano@gentoo.org> emerge, portage.py: Fixed 2 bugs.
- Portage doesn't read local virtuals file, which happens on only cvs
- version. package is blocked by itself.
-
- 21 Jan 2004; Nicholas Jones <carpaski@gentoo.org> portage.py, emerge:
- Fix from genone for emerge's direct reading of packages and his patch
- that also adds in /etc/portage/profile as a stacked profile.
-
- 21 Jan 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Completed
- inheritence capabilities for portage.config reading some files. Moved a
- copy of the getvirtuals() function into settings to handle multiple
- profiles properly.
-
-*portage-2.0.50_pre17/18/19 (21 Jan 2004): Modules for DBs and quick fixes
-
- 21 Jan 2004; Nicholas Jones <carpaski@gentoo.org> *: Moved all references
- to var/db/pkg to portage.VDB_PATH --- This will change again -- NEED TO
- BE MOVED INTO A PATH/CONSTANTS SETUP.
-
- 21 Jan 2004; Nicholas Jones <carpaski@gentoo.org> dosed: Quick fix for
- the basename missing/misplaced issue.
-
- 21 Jan 2004; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Added
- /dev/console to PREDICT to attempt a workaround for a serial console
- bug. dbkey is now set through portage.py/doebuild to allow for modular
- db code.
-
- 21 Jan 2004; Nicholas Jones <carpaski@gentoo.org> emerge: regen doesn't
- require root anymore. Edited the timestamp check to be a little more
- friendly -- delete the portdir timestamp and it won't use the alternate.
- Fix some permission settings. Added some warnings in for cachedirs that
- are very likely to ruin your system. Cleaned out some of the eclass code
- that isnt valid any longer.
-
- 21 Jan 2004; Nicholas Jones <carpaski@gentoo.org> portageq: Added vdb_path
- as a target to get the db directory. Quickpkg uses this.
-
- 21 Jan 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Added
- load_mod() -- grabs a class/function from a module and passes it back
- without loading the module into the global scope. Added unique_array()
- which eliminates duplicates from an array. grab_stacked() operates like
- the other grab* and getconfig functions, but takes a filename and a set
- of paths that it will apply incrementally or clobbers -- for profile
- inheritance. getconfig no longer exits on non-existance returns None.
- Class config now should be passed a profile path and a set of incremental
- values instead of using the globals -- defaults to using the globals
- presently and print an error message. Adding support for module configs
- as a set of strings 'class.subclass.objectmodule':'module.to.use.object'
- for load_mod and the database modules. Profile inheritance started. Killed
- the eclass() super-function and replaced it with class eclass_cache that
- is visible and conceptually simpler -- Also uses the plugable modules.
- Cleaned out the sync calls for the DBs. MASSIVE simplification of the
- aux_get code -- removed memory-caching in favor of system cache (actually
- faster in all cases so far -- P100 and P4-2.2G). Lockfile usage around the
- cachefile.
-
- 21 Jan 2004; Nicholas Jones <carpaski@gentoo.org> portage_db_*: Updated
- the API a little but to have permissions set properly. A little more
- reorganization and removed the keycount checks.
-
- 21 Jan 2004; Masatomo Nakano <nakano@gentoo.org> emerge: download size
- should not be displayed when the package is nomerge with --tree.
-
- 20 Jan 2004; Masatomo Nakano <nakano@gentoo.org> portage.py: Reverted
- ambiguity package fix in cpv_expand().
-
- 20 Jan 2004; Masatomo Nakano <nakano@gentoo.org> emerge: Moved
- backup timestamp.chk file from portage tree to PORTAGE_TMPDIR.
-
- 20 Jan 2004; Masatomo Nakano <nakano@gentoo.org> emerge: Reverted the
- backing up the timestamp.chk fix.
-
- 20 Jan 2004; Masatomo Nakano <nakano@gentoo.org> emerge: Improved the
- list of --tree by TGL's patch. This should close #38070.
-
- 20 Jan 2004; Masatomo Nakano <nakano@gentoo.org> emerge: Removed debug
- message without --debug. This should close #23840.
-
- 19 Jan 2004; Masatomo Nakano <nakano@gentoo.org> emerge: Improved
- timestamp check of rsync. This should close #37403.
-
- 19 Jan 2004; Masatomo Nakano <nakano@gentoo.org> portage.py: Improved
- regeneration ld.so.cache. This should close #37858.
-
- 19 Jan 2004; Masatomo Nakano <nakano@gentoo.org> emerge: Fixed bug which
- emerge doesn't block same package but different version.
- (example: DEPEND="!<cat/pkg-1.0.0" in cat/pkg-1.0.0.ebuild)
-
- 19 Jan 2004; Masatomo Nakano <nakano@gentoo.org> portage.py:
- Modified cpv_expand() to check package.mask. This should close #38592.
-
- 19 Jan 2004; Masatomo Nakano <nakano@gentoo.org> portage.py: Escaped
- regualar expression for replace entry in fixdbentries().
-
- 18 Jan 2004; Masatomo Nakano <nakano@gentoo.org> portage.py:
- Fixed AUTOCLEAN delay problem in .50pre* by TGL's patch. This close
- #38189. Fixed unmerge failture bug when 'ebuild foo-1.0.0 unmerge'.
- These close #38189, #38366
-
- 18 Jan 2004; Masatomo Nakano <nakano@gentoo.org> emerge, portage.py:
- Fixed "ebuild /foo/bar-1.0.0.ebuild unmerge" and "emerge bar-1.0.0 unmerge"
- problems. This should close #38420.
-
- 17 Jan 2004; Masatomo Nakano <nakano@gentoo.org> portage.py: Fixed
- "!<=" style block problem. Fixed symlink with absolute path
- problem in treewalk().
-
-*portage-2.0.50_pre16 (13 Jan 2004): Quick Fixes -- ~arch version
-
- 13 Jan 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Removed an
- unnecessary depend call that double eclass-using ebuild's cache regen
- time.
-
-*portage-2.0.50_pre15 (12 Jan 2004): Quick Fixes -- ~arch version
-
- 12 Jan 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Unmerge
- traceback fix.
-
-*portage-2.0.50_pre14 (12 Jan 2004): Quick Fixes -- ~arch version
-
- 12 Jan 2004; Nicholas Jones <carpaski@gentoo.org> emerge: Fix for
- traceback on '-S'.
-
- 12 Jan 2004; Nicholas Jones <carpaski@gentoo.org> repoman: Fix for
- traceback on --help.
-
- 12 Jan 2004; Nicholas Jones <carpaski@gentoo.org> sandbox: Fix for
- sandboxpids.tmp file accesses.
-
- 12 Jan 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Sandbox, as
- above. Catch invalid package names and print a sane message about it.
-
-*portage-2.0.50_pre13 (11 Jan 2004): Fixes
-
- 11 Jan 2004; Nicholas Jones <carpaski@gentoo.org> cnf/*: Updated the
- Advanced masking section to aid the reduction of user complaints and
- requests for unreasable usage of ACCEPT_KEYWORDS.
-
- 11 Jan 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: A counter
- fix was fixed to actually check the counters of all CP versions to ensure
- the new counter is higher than all existing ones. Modified the dblink
- class to have class lockfiles for the db and tmpdb dirs as well as lock
- other files before editing. Reorganization of the merge code in dblink
- so that the tmpdb is filled immediately after preinst and prior to the
- actual FS merging -- COUNTER and CONTENTS go directly into the tmpdb
- and not into the infodir.
-
-*portage-2.0.50_pre11/12 (09 Dec 2003): repoman/binpkg/exit conditions
-
- 09 Jan 2004; Nicholas Jones <carpaski@gentoo.org> emerge: getbinpkgonly
- fixes for emerge -G world, should behave properly now instead of using
- ebuild masks. Only downloads immediately before a merge -- fetchonly now
- applies to binary packages.
-
- 08 Jan 2004; Masatomo Nakano <nakano@gentoo.org> repoman: Ignore other
- arches check in repoman when --ignore-other-arches(-I).
-
-*portage-2.0.50_pre10 (06 Dec 2003): API change + enhancements
-
- 06 Jan 2004; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Fix for
- dyn_preinst being called before IMAGE was set -- IMAGE is now valid
- in pkg_preinst. Added suidctl for SELinux.
-
- 06 Jan 2004; Nicholas Jones <carpaski@gentoo.org> emerge: Added -P to
- initial cvs checkout.
-
- 06 Jan 2004; Nicholas Jones <carpaski@gentoo.org> quickpkg: Fix for
- the 'tar up /' problem.
-
- 06 Jan 2004; Nicholas Jones <carpaski@gentoo.org> portage.py: Caught a
- traceback generated by bad depend atoms for repoman. Fixes from genone
- for package.*. Fixed the checks for doebuild calls in treewalk that was
- ignoring exit conditions for ebuilds.
-
- 04 Jan 2004; Masatomo Nakano <nakano@gentoo.org> repoman: Added PDEPEND
- dependency check. This closes #24796
-
- 04 Jan 2004; Masatomo Nakano <nakano@gentoo.org> repoman, portage.py:
- Added new dependency check to repoman. This closes #36887.
-
- 03 Jan 2004; Masatomo Nakano <nakano@gentoo.org> emerge: Modified
- to specific port number in emerge sync. This closes #36994
-
- 02 Jan 2004; Masatomo Nakano <nakano@gentoo.org> portage.py: Fixed
- a problem that emerge doesn't block package when it's required.
- It happens in .50_pre*.
-
- 02 Jan 2004; Masatomo Nakano <nakano@gentoo.org> portage.py: Fixed
- issue with getsize() when --debug.
-
- 02 Jan 2004; Masatomo Nakano <nakano@gentoo.org> portage.py: Fixed
- issue with virtual. This closes bug #9050, #22225, #29499.
-
- 01 Jan 2004; Masatomo Nakano <nakano@gentoo.org> ebuild, emerge, portage.py:
- Fixed issue with not cleaning up temp directory. This closes bug #34967.
-
- 31 Dec 2003; Masatomo Nakano <nakano@gentoo.org> emerge:
- Fixed 'emerge sync' issue which continuously connects to same host.
-
- 31 Dec 2003; Nicholas Jones <carpaski@gentoo.org> emerge: Found the line
- that was causing the package dir to be printed... It was a spawn call.
-
- 31 Dec 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Fix for
- the symlink corruption in the db from the movefile() bug.
-
- 29 Dec 2003; Masatomo Nakano <nakano@gentoo.org> portage.py:
- Fixed bug which emerge stops when no denpendencies exist in || ( )
- by USE flags. This closes #36568.
-
- 29 Dec 2003; Masatomo Nakano <nakano@gentoo.org> emerge, portage.py:
- Added an ambiguity package check when emerge. This closes bug #22700.
-
-*portage-2.0.50_pre9 (24 Dec 2003): API change + enhancements
-
- 24 Dec 2003; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Added
- PORTAGE_TMPDIR to SANDBOX_READ/WRITE to ensure it works. SpanKY's
- patch for use negation added (use !foo). pkg_setup doesn't die on
- a non-zero exit status.
-
- 24 Dec 2003; Nicholas Jones <carpaski@gentoo.org> emerge: using os.uname
- instead of calling out to uname.
-
- 24 Dec 2003; Nicholas Jones <carpaski@gentoo.org> quickpkg: Added SpanKY's
- patch for delayed exit/error conditions.
-
- 24 Dec 2003; Nicholas Jones <carpaski@gentoo.org> xpak.py: chdir's added
- to the getcwd fix for missing dirs.
-
- 24 Dec 2003; Masatomo Nakano <nakano@gentoo.org> emerge: Added OVERLAY
- directories display for --verbose.
-
-*portage-2.0.50_pre8 (24 Dec 2003): API change + enhancements
-
- 22 Dec 2003; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Added
- /proc/self/maps to SANDBOX_PREDICT, and /dev/shm to read/write.
-
- 22 Dec 2003; Nicholas Jones <carpaski@gentoo.org> emerge: Added automake
- and autoconf versions to the output of emerge info.
-
- 22 Dec 2003; Nicholas Jones <carpaski@gentoo.org> etc-update: Added
- edit merged file option -- defaults to EDITOR var or "nano -w".
-
- 22 Dec 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Use
- os.uname instead of calling out to uname which might not exist.
-
-*portage-2.0.50_pre7 (22 Dec 2003): API change + enhancements
-
- 22 Dec 2003; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: patch to
- quote most of the path operators that might involve spaces.
-
- 22 Dec 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Fix for
- invalid entries in package.keywords. Character chopping on mirrors
- fixed again.
-
- 21 Dec 2003; Masatomo Nakano <nakano@gentoo.org> bin/ebuild, bin/emerge,
- pym/portage.py: Changed to show disabled USE flags from use.mask when
- using emerge -vp. And fixed use.mask issue.
-
- 20 Dec 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Rewrote
- match_from_list -- Simplified and made pkgcmp and match_from_list
- properly compare package names.
-
- 20 Dec 2003; Nicholas Jones <carpaski@gentoo.org> repoman: Fix for mysigs
- traceback when signing.
-
- 20 Dec 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Added
- PYTHONPATH to the specials list -- created a colon_seperated list.
- Fixed reset() in class config so that you can specify keeping the
- pkg dictionary when resetting the values.
-
- 19 Dec 2003; Masatomo Nakano <nakano@gentoo.org> repoman: Added check
- whether "ebuild foo.ebuild digest" succeeds.
-
- 19 Dec 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Fix for
- pkg settings being maintained after an unmerge.
-
- 19 Dec 2003; Nicholas Jones <carpaski@gentoo.org> pym/portage_db_*: Moved
- to using cPickle instead of marshal. More standardization of the API.
-
- 18 Dec 2003; Masatomo Nakano <nakano@gentoo.org> repoman: Added virtual
- dependency check on each arch.
-
- 17 Dec 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Fixed a
- permission issue involving $T and userpriv. Lockfile touchup.
-
- 17 Dec 2003; Nicholas Jones <carpaski@gentoo.org> portage_db_*: Added
- templates and db for cache interfaces. Presently have a anydbm and a
- flat file interface working. See the test for operations.
-
- 15 Dec 2003; Nicholas Jones <carpaski@gentoo.org> emerge: Added a call
- to portageq that causes python to create optimized modules prior to it
- ending up inside the sandbox. Added more output and logging to sync.
-
- 15 Dec 2003; Nicholas Jones <carpaski@gentoo.org> prepstrip: 'tree' is not
- the same as 'true'.
-
- 15 Dec 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: invalid
- settings in package.keywords caused a traceback -- fixed with error message.
-
-*portage-2.0.50_pre1 (12 Dec 2003): API change + enhancements
-
- 10 Dec 2003; Nicholas Jones <carpaski@gentoo.org> chkcontents: Uses portage
- functions to do md5sum calcs.
-
- 10 Dec 2003; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Removed try()
- as it isn't used, and was deprecated for a long while. Genone's fetching
- size display added for --verbose. License display added. Added a little
- debug for IUSE so we can figure out the binary package --verbose IUSE
- issues that are randomly reported. XXXXXXXXXXXXXXXXXXX's 'buildsyspkg'
- patch for building only system packages into tbz2s. Unmerge fix for new
- settings instances. RSYNC_RATELIMIT added.
-
- 10 Dec 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: ADA path
- variables added to specials for env_update. Error messaeg correction for
- make.defaults syntax errors. Unmerge now uses the environment file, if it
- exists, to get the complete environment back to perform unmerge operations.
- load_infodir() uses pkg settings completely now. Fixed the passing of
- settings in unmerge and dblink. Fixed an issue regarding unlinking lockfiles
- while inside of a sandbox.
-
- 09 Dec 2003; Nicholas Jones <carpaski@gentoo.org> ebuild.sh, *.sh:
- Moved helper scripts into bin/functions and made them sourceable -- they
- now will die in cases where sub-parts fail. dodoc and keepdir are now
- recursive-capable.
-
- 09 Dec 2003; Nicholas Jones <carpaski@gentoo.org> emerge: emerge.log now
- set as portage:portage with 0660 perms. --debug now enables tracebacks
- for dep generation instead of moving code out of the try block.
-
- 09 Dec 2003; Nicholas Jones <carpaski@gentoo.org> g-cpan.pl: rac's patch
- to get arch list from portage's list of arches in the profiles.
-
- 09 Dec 2003; Nicholas Jones <carpaski@gentoo.org> repoman: Moved a bit of
- the existing gpg code around -- it might work as is, but requires 'sign'
- in features. Fixed a potential for repoman to miss updates that should
- get a new manifest and commit. Fixed digest/manifest generation for
- non-packagedir runs of repoman.
-
- 09 Dec 2003; Nicholas Jones <carpaski@gentoo.org> emergehelp.py, make.conf,
- getbinpkg.py: Message touch ups.
-
- 09 Dec 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: load_infodir()
- uses pkg settings now instead of env and backup. Genone's custom mirror
- patch included. Added some missing 'strict' flags for recursion in digest*().
- Refixed the invalidentry stuff that was lost across patch merges. Fix for
- pkg-keywords from genone included. Genone's deprecated profile patch for
- reporting to a user that their current profile is deprecated. Message about
- missing arch.list instead of spouting invalid keywords messages.
-
- 08 Dec 2003; Masatomo Nakano <nakano@gentoo.org> repoman:
- Added all arch dependency check. This closes bug #24160.
-
- 07 Dec 2003; Masatomo Nakano <nakano@gentoo.org> emerge,portage.py:
- Fixed bugs. 1.--debug doesn't work 2.Portage breaks files
- in /var/db/*/*. 3.No stop if dependency problem happens.
- They are only cvs version problems.
-
- 01 Dec 2003; Masatomo Nakano <nakano@gentoo.org> emerge: Fixed bug which
- always remakes info dir file.
-
- 29 Nov 2003; Masatomo Nakano <nakano@gentoo.org> portage.py: Fixed issue with
- ebuild name rule. Fixed typo with variable name.
- This closes bug #17172,#34666
-
- 29 Nov 2003; Masatomo Nakano <nakano@gentoo.org> emerge: Fixed issue with
- lacking the "setting" argument for pkgmerge()
-
- 29 Nov 2003; Masatomo Nakano <nakano@gentoo.org> emerge: fixed rsync bug.
- This closes bug #34660.
-
- 28 Nov 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Migration
- to non-global settings started -- class config has new functionality and
- is locked after portage is finished initializing -- changes cannot be made
- to the global instance of config 'settings' -- reset() is now functional,
- setcpv() loads PKGUSE from /etc/portage/package.use, load_infodir() loads
- all small files (under 4k) from the vardb directory of an installed package
- so that operations have the same post* settings as they had at merge time.
- Begin modifications to spawn() to allow for files/pipes to be used for
- IO instead of using getstatusoutput which does not take an environment
- parameter like execve(). check_config_instance() ensures that the provided
- parameter is a 'class config' instance -- for ensuring that everything is
- being passed properly with the changes. Fix for the local FS mirror issue
- where it removed the first '/' instead of the last one. doebuild() cleanups
- for readability and pkguse enhancements -- also remove getstatusoutput()
- usage for depend so that we don't have to modify the active environment.
- Fix for symlink mtime values returned from movefile. (Nakano) SLOTMOVE
- added to global update functionality to fix some issues where a package
- suddenly must become slotted. portdbapi takes a root parameter instead
- of using settings. Slightly more useful output from depend. binarytree()
- now takes a pkgdir instead of using settings. Portage will now die if
- ebuild.sh exits on a signal.
-
- Moved some functions around and renamed them for general use -- derived
- from match2 in class portagetree:
- match_to_list() find all atoms in a list that match a given package.
- best_match_to_list() determines the most specific match. Needs work.
- match_from_list() find all packages in a list that match a given atom.
-
- 28 Nov 2003; Nicholas Jones <carpaski@gentoo.org> emerge: Fixed an issue
- with searchdesc wanting root permissions if run as non-root. Migrated to
- the non-global config class. EMERGE_FROM added for the dyn_preinst patch
- -- Indicates if a merge is occuring from an ebuild or from a binary. Patch
- for rsync timestamp checking from Nakano.
-
- 28 Nov 2003; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Save PKGUSE.
- Pebeneto's patch for dyn_preinst and SELinux added as a fix for binary
- and ebuild merges. Added a kill for portage during the depend phase so
- that portage will actually die if you control-C.
-
- 28 Nov 2003; Nicholas Jones <carpaski@gentoo.org> *: MASSIVE set of changes
- to start using locally defined 'class config' instances. This allows us to
- start working on some parallelism among other things. Created this way:
- mysettings = portage.config(clone=portage.settings)
-
- The Following functions now take a 'config' parameter:
- spawn(), fetch(), digestgen(), digestcheck(), spawnebuild(), doebuild(),
- merge(), dep_opconvert(), dep_check(), dblink.__init__()
-
- package.keywords is now implemented curtasy of genone/max. PKGUSE was
- rewritten for the global config killing and is also included. X11 man
- pages now found and zipped correctly. SYS.PATH fixes for the python
- migration -- issue actually only shows up on 2.2 systems because of how
- compiled modules are used if found regardless of the original source's
- existance.
-
- 28 Nov 2003; Nicholas Jones <carpaski@gentoo.org> tabcheck.py: An easier
- way to make sure that all the python stuff is correctly using tabs and
- not mixing spaces.
-
- 28 Nov 2003; Nicholas Jones <carpaski@gentoo.org> xpak, xpak.py: Fixes
- to ensure that it works if the current dir is missing and that the python
- path gets set properly.
-
- 22 Nov 2003; Daniel Robbins <drobbins@gentoo.org> portage.py: Fixed
- calls in vartree method to invalidentry().... made them call call
- self.dbapi.invalidentry() (there were multiple wrong method calls.)
-
- 10 Nov 2003; Nicholas Jones <carpaski@gentoo.org> md5check.py: Checks all
- digests and SRC_URIs for filenames and associated MD5s. Reports collisions
- between versions/packages, missing, and extra lines in digests.
-
-*portage-2.0.49-r17/18 (10 Nov 2003): Fixes
-
- 10 Nov 2003; Nicholas Jones <carpaski@gentoo.org> *: Changed portage to
- be the first path in sys.path for all python scripts. Also enabled
- optimizations from the scripts to ensure everything imported is built
- for speed. ebuild: applied fix for the '//' root breaking the db[].
- prepstrip: etdyn quickfix
-
- 10 Nov 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: fix for
- RESTRICT=nouserpriv. GENTOO_MIRRORS can have paths set to take files
- from. Fixes for mishandled cache data regarding *pkgsplit(). Fixes for
- '*' being returned as part of a package split. An 'invalidentry()' fix
- for a traceback. Nakano's fixes for virtual removals not working properly,
- sandbox violations during pkg_nofetch, || depend selection. Genone's
- fixpackages speedup.
-
- 10 Nov 2003; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: PORTAGE_TMPDIR
- fix for distcc. Variable passing bug patch for export_functions. Exit 1
- added for nofetch to stop sandbox violation. Nakano's --tree patch added.
- Improved the unmerge messages to denote what kind of unmerge fails. Info
- pages regex pattern adjusted to allow most any name for a page.
-
- 10 Nov 2003; Nicholas Jones <carpaski@gentoo.org> repoman: genone's xml
- linting additions.
-
- 01 Nov 2003; Robin H. Johnson <robbat2@gentoo.org> pym/cvstree.py:
- fix bug #32071, by properly escaping a string to not be a regex. Checked
- thru entire *.py tree and found this is the only mis-use of strings that
- need to be escaped.
-
- 31 Oct 2003; Daniel Robbins <drobbins@gentoo.org> portage.py: /lib/modules
- now gets "unmerge protection." This is half of the config protection
- functionality. It means that anything in /lib/modules will not be deleted
- when a package is unmerged (often automatically when a user merges a
- kernel module ebuild for a new kernel.) This solves the "my module
- disappeared!" issue. This closes bug #1477.
-
- 31 Oct 2003; Daniel Robbins <drobbins@gentoo.org> emerge: Should no longer
- spit out wacky "!!! no match found" warnings when auto-cleaning.
-
- 30 Oct 2003; Daniel Robbins <drobbins@gentoo.org> portage.py: Only run
- depscan.sh if it exists on disk. This allows Portage to run inside a stage1
- where /sbin/depscan.sh doesn't exist.
-
- 30 Oct 2003; Daniel Robbins <drobbins@gentoo.org> portage.py: Applied fix to
- allow multi-level "use? ( )" in SRC_URI, closing bug #16159.
-
-*portage-2.0.49-r15/16 (21 Oct 2003): Fixes
-
- 21 Oct 2003; Nicholas Jones <carpaski@gentoo.org> fix-db.py: was broken
- for python2.3 -- fixed now.
-
- 21 Oct 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Added
- lockfiles to prelink md5 checks. Fixed caching bug where cache objects
- were passed back as pointers instead of copies. Added 'invalidentry'
- function to handle lockfiles -- It tests/deletes them using unlockfile.
- Added fix-db.py to the 'databases is broken' messages.
-
- 21 Oct 2003; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: added CDPATH
- to unset. SELinux fix for sandbox.
-
-*portage-2.0.49-r13/14 (10 Oct 2003): Fixes
-
- 10 Oct 2003; Nicholas Jones <carpaski@gentoo.org> *: Full adaptations
- for python2.3 implemented. Installation setup for /usr/lib/portage/pym
- instead of site-packages. Fix for Old-Instance unmerging which stopped
- happening due to changed path names. Binaries shouldn't merge under
- fetchonly.
-
-*portage-2.0.49-r11/12 (08 Oct 2003): Internal Only.
-
-*portage-2.0.49-r10 (08 Oct 2003): Fixes
-
- 08 Oct 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Fixed a
- seemingly random traceback involving lockfiles -- Categories weren't
- being created before a lockfile was attempted in the category directory.
- More enhancements to aid migration to python-2.3. Added writemsg() as
- a general function for outputting information -- Takes an argument that
- is interpreted as debug level and prints synchronisly to stderr. Yanked
- domenu pending a GLEP.
-
-*portage-2.0.49-r9 (07 Oct 2003): Fixes
-
- 07 Oct 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Atomic lock
- updates and self.create() fixes. Counter enhancements.
-
-*portage-2.0.49-r8 (05 Oct 2003):
-
- 05 Oct 2003; Nicholas Jones <carpaski@gentoo.org> *: Changed #! line to
- use /usr/bin/python to aid in migration.
-
- 05 Oct 2003; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: More use/has
- output fixups.
-
- 05 Oct 2003; Nicholas Jones <carpaski@gentoo.org> emerge: General cleanups.
- Added in baselayout info to emerge info. --skipfirst isn't a persistent
- option on resume anymore.
-
- 05 Oct 2003; Nicholas Jones <carpaski@gentoo.org> fix-db.py: New script to
- aid in diagnosing and eventually fixing /var/db issues. It makes almost no
- modifications at the moment.
-
- 05 Oct 2003; Nicholas Jones <carpaski@gentoo.org> prepstrip: Addition of
- a condition for etdyn binaries -- they list as shared objects but aren't.
-
- 05 Oct 2003; Nicholas Jones <carpaski@gentoo.org> repoman: Added non-cvs
- patch -- repoman can be used outside of cvs trees for scanning/checking.
- Added a masking fix for packages that use 'arch?' dependencies. Permission
- fix on stats pickle.
-
- 05 Oct 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Subversion
- directories are ignored along with CVS in listdir() when ignorecvs is set.
- USE_EXPAND is fixed. Global and package counter fixes. ATOMIC VARDB moves
- are now implemented -- Still need a way to recover broken operations.
- Counter functions are now implemented in dbapi only. vardbapi now implements
- aux_get for all possible files listed in the package's info directory. New
- functions: lockfile()/lockdir(), unlockfile()/unlockdir().
-
-*portage-2.0.49-r7 (26 Sep 2003): Binary package fixes + spacing issues.
-
- 26 Sep 2003; Nicholas Jones <carpaski@gentoo.org> emerge.sh: Spacing fixes
- for tab/space mixes. glob fix for getgccversion().
-
- 26 Sep 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Fix for binary
- use flag tb.
-
- 26 Sep 2003; Nicholas Jones <carpaski@gentoo.org> xpak.py: Fix for scan()
- tb when file is invalid or shorter than XPAK header length.
-
-*portage-2.0.49-r6 (23 Sep 2003): SELinux, DistCC, and pretend output fixes.
-
- 23 Sep 2003; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: selinux context
- fixes -- using ${T} now. DISTCC_DIR added by default and given an addwrite.
-
- 23 Sep 2003; Nicholas Jones <carpaski@gentoo.org> emerge: gccversion() added
- to help distcc and the version checking calls that break distcc permissions.
- Fix for create() and the useflag passing from binary packages. When using
- pkgs, is_newer_ver_installed() no longer trashes ebuild names. Don't look
- for fetch restrictions with binary packages. Included a modified patch that
- shows only in-slot versions for packages during pretend output -- In-slot
- now shows up exactly as a single-slot package would -- New slots show up
- as new packages -- proper output. Killed debug for everything except regen
- which now shows the deps being regenerated. Fixed regen to ignore keywords
- just like sync does already... should fix a few missing cache entry problems
- on rsync1 and speed up sync times all around.
-
- 23 Sep 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Additions for
- distcc support.
-
-
-*portage-2.0.49-r5 (19 Sep 2003): Fixes
-
- 19 Sep 2003; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Make
- DISTCC_DIR set if not defined.
-
- 19 Sep 2003; Nicholas Jones <carpaski@gentoo.org> emerge: Fix for -U
- introduced bugs with Str+None tracebacks. Mild change to log info -- Added
- short pkgname before ebuild. Eclassdb changes and flush/save calls.
-
- 19 Sep 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Unregister
- signal handling after receiving the signal and resend the signal we got.
- Run depscan during env-update. Migrated eclass stuff into it's own pickle.
- Code for moving /var/cache/edb/dep into a pickle for each category added --
- disabled via hardcoded value presently. Removed some try blocks in favor
- of detecting the cause of an error first -- Hopefully gives a little speed
- up. Cache fixes and useful output added. A tbz2 moving bug where an error
- was printed about files the destination existing already is now fixed. The
- dircache is cleared on every unmerge now, to ensure the cache does not
- interfere. bzip2 is spawned with the quiet flag now. Update list is sorted
- for proper year/quarter order now. -arch isn't an invalid keywork anymore.
- Generic pickle reading and writing functions added.
-
-*portage-2.0.49-r4 (10 Sep 2003): Fixes
-
- 10 Sep 2003; Nicholas Jones <carpaski@gentoo.org> *: Added the facility
- to incorporate binary package use flags when calculating deps. Changes in
- -r2 and -r3 included quick fixes to SELinux code and the addition of
- LINGUAS to USE_EXPAND.
-
-*portage-2.0.49-r1 (25 Aug 2003): Fixes
-
- 25 Aug 2003; Nicholas Jones <carpaski@gentoo.org> emerge: -K traceback fix.
- Made fetchonly quit traversing the merge code after fetching.
-
- 25 Aug 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Removed
- auxcache saving code -- A few bugs and caching problems need to be resolved.
- Moved a block of code in aux_get inside of an existing conditional -- Should
- provide a small speedup.
-
-*portage-2.0.49 (22 Aug 2003): GRP downloading, Selinux, General Fixes
-
- 22 Aug 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Make
- auxcache saving more friendly.
-
-*portage-2.0.49_pre20/21 (20 Aug 2003): Fixes + security enhancements
-
- 20 Aug 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Fix for
- auxcache saving.
-
-*portage-2.0.49_pre19 (20 Aug 2003): Fixes + security enhancements
-
- 20 Aug 2003; Nicholas Jones <carpaski@gentoo.org> *: Updates to Wayne's
- modifications on dispatch-conf.
-
- 20 Aug 2003; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: sfperms
- added to strip permissions if set in features. selinux context support.
- Nakano's cleanups for unmerge output. Added distcc and ccache versions
- into emerge info's output. 'autoaddcvs' FEATURE is documented in make.conf.
-
- 20 Aug 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Selinux code
- added. No longer traceback on cp_list for categories. Added in a patch to
- use a pickle for auxcache storing. Should help slow IO boxes.
-
-*portage-2.0.49_pre18 (15 Aug 2003): Fixes.
-
- 15 Aug 2003; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Nakano's
- fix for a glob expansion bug on a tar call.
-
- 15 Aug 2003; Nicholas Jones <carpaski@gentoo.org> emerge: Nakano's fixes
- for exiting on a signal and incorrect output for the unmerge screen.
-
- 15 Aug 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Added in
- patch to shuffle thirdparty mirrors. Nakano's fixes for digests+fetching
- bugs/messages/errors. Virtual stripping fix.
-
- 15 Aug 2003; Nicholas Jones <carpaski@gentoo.org> *: Fix to emerge -V
- and related output. Typo fixes. Parent thread no longer drops root in
- userpriv. 'autoaddcvs' feature now determines if portage will auto-add
- files to cvs. Virtuals fix for an empty key line. Added user categories
- file as /etc/portage/categories.
-
-*portage-2.0.49_pre17 (30 Jul 2003): Resuming/wget GRP, Sandbox updates
-
- 30 Jul 2003; Nicholas Jones <carpaski@gentoo.org> *: GRP updates to
- use RESUMECOMMAND to determine how to download GRP binaries. Adjustments
- to the portage.spawn() code to try and speed it up by eliminating the
- copy phase (copy-on-write forking). Massive messages for corrupt FS issues.
- Movefile() checks to ensure we actually complete the moves. Killed the
- broken pipe message for tbz2 extraction.
-
-*portage-2.0.49_pre16 (29 Jul 2003): Touchups
-
- 29 Jul 2003; Nicholas Jones <carpaski@gentoo.org> *: Clean up the debug
- output and quickfix downloading tbz2s.
-
-*portage-2.0.49_pre15 (29 Jul 2003): Fixes
-
- 29 Jul 2003; Nicholas Jones <carpaski@gentoo.org> repoman: imported time.
-
- 29 Jul 2003; Nicholas Jones <carpaski@gentoo.org> emerge: Corrected the
- logic used to determine whether a package or an ebuild is used when given
- the option.
-
- 29 Jul 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Changed a
- lot of calls using 'print' into sys.stderr.write() calls. Added try/except
- around the source lstat to catch funky errors where the file doesn't really
- exist -- Presents descriptions to users about what to do. Added more catches
- for movefile() calls to ensure it dies on failures. Changed the tbz2
- extraction call to stop the 'cat: broken pipe' message.
-
-*portage-2.0.49_pre14/48-r7 (24 Jul 2003): Fixes and getbinpkg caching
-
- 22 Jul 2003; Nicholas Jones <carpaski@gentoo.org> *: Typo fixes for
- the ECLASS_DEPTH comparisons in ebuild.sh. Traceback fixes for blocking
- packages in depclean and in pretend. Caching added to getbinpkg code --
- generation of cache supported, but cachefile is staticly located. Langs
- patch from Nakano in bug #9988 included.
-
-*portage-2.0.49_pre13 (22 Jul 2003): Fixes
-
- 22 Jul 2003; Nicholas Jones <carpaski@gentoo.org> doman: Doesn't gzip
- .keep files now.
-
- 22 Jul 2003; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: USERLAND
- fix. ** ECLASS depth tracker and additions to the new depend code. **
-
- 22 Jul 2003; Nicholas Jones <carpaski@gentoo.org> emerge: Fetch restriction
- now adds a red F to the pretend output. Made the blockers message better.
- Added 'local' to rsync excludes to allow the category to be added for admins.
-
- 22 Jul 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Fixed a TB
- caused by a missing 'strict' reference --- Needs to be fixed better. Fixed
- a typo in the populate code for bintree/getbinpkg.
-
-*portage-2.0.49_pre11 (16 Jul 2003): GRP, General, VIDEO_CARDS/INPUT_DEVICES
-
- 16 Jul 2003; Nicholas Jones <carpaski@gentoo.org> quickpkg: SpanKY's
- friendly edition now included.
-
- 16 Jul 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: USE_EXPAND
- support for expanding bash variables into USE --- VIDEO_CARDS="blah"
- USE="video_cards_blah". Added a notice about fetching metadata so you
- can see that it's happening.
-
- 16 Jul 2003; Nicholas Jones <carpaski@gentoo.org> make.globals*: Added
- default CHUNKSIZE for binhosts. Added defaults for USE_EXPAND which is
- used to expand variables into USE from the give names.
-
- 16 Jul 2003; Nicholas Jones <carpaski@gentoo.org> *: Debug removal.
-
-*portage-2.0.49_pre10 (16 Jul 2003): GRP and General Fixes
-
- 16 Jul 2003; Nicholas Jones <carpaski@gentoo.org> *:
- Quick fix for a breakage in the GRP --getbinpkg code.
-
-*portage-2.0.49_pre9 (16 Jul 2003): GRP and General Fixes
-
- 16 Jul 2003; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: add
- LDFLAGS and ASFLAGS. Exports are only done if vars are already set.
-
- 16 Jul 2003; Nicholas Jones <carpaski@gentoo.org> emerge: Added support for
- --getbinpkg and --getbinpkgonly --- GRP complement. Fixes for blocking-
- not-working bug
-
- 16 Jul 2003; Nicholas Jones <carpaski@gentoo.org> repoman: Patch to detect
- incomplete digests.
-
- 16 Jul 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: ebuild fetch
- now checks md5sums (run from ebuild). Support for getbinpkg.
-
- 16 Jul 2003; Nicholas Jones <carpaski@gentoo.org> xpak.py: Added more in-
- memory operations support.
-
- 16 Jul 2003; Nicholas Jones <carpaski@gentoo.org> getbinpkg.py: Supporting
- code for binary package retrieval.
-
-*portage-2.0.48-r2 (29 Jun 2003): Fixes and Multiple Overlays
-
- 29 Jun 2003; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Fixed the
- commands for tar.
-
- 29 Jun 2003; Nicholas Jones <carpaski@gentoo.org> emerge: Mode the chdir
- to fix the spanky bugs down to post_emerge to fix the "can't merge an
- ebuild file with an absolute path" bug.
-
- 29 Jun 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Added function
- suffix_array(array,suffix) which takes an array and adds suffix to the end
- of each element. Added 'cd / ;' into the getstatusoutput() calls to ldconfig.
- Added a check so that portage doesn't try to add files/dir from a non-cvs
- directory. Added 'manifest' target to ebuild/portage to only adjust the
- manifest. Added Multiple overlay support -- Should be fairly thorough --
- needs more debug though. Fixed portdbapi::new_protect_filename().
-
-*portage-2.0.48-r2 (29 Jun 2003): Cleanups and Fixes (testing)
-
- 29 Jun 2003; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Lots of
- consistency cleanups. Added more files to the var/db data for future
- use. Added more die conditions to failure points. Fixed/Hacked a fix
- into the inherit-not-dieing problem. Set TMPDIR and TMP globally. Added
- notice for multiple inheritance. Fixed inheritance problem with multiple
- inheritance clobbering previous list of eclasses. Basic support for
- multiple overlays added to ebuild.sh.
-
- 29 Jun 2003; Nicholas Jones <carpaski@gentoo.org> emerge: SRC_URI fix for
- 'rm /etc/*' problem. Spanky bug: chdir to / to avoid problems when merging
- from inside a builddir that gets deleted.
-
- 29 Jun 2003; Nicholas Jones <carpaski@gentoo.org> etc-update: Added notice
- for symlinks in cfg_prot setups.
-
- 29 Jun 2003; Nicholas Jones <carpaski@gentoo.org> repoman: complain if
- FEATURES=cvs not set.
-
- 29 Jun 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Added in
- a work around for the 'USE=+useflag' users. Added support for nouserpriv
- via RESTRICT. Fix the bug with ROOTs with no previous file causing a TB
- from counter_tick(). Make symlinks follow config_protect just like normal
- files.
-
- 10 Jun 2003; Daniel Robbins <drobbins@gentoo.org> ebuild.sh, portage.py,
- various commands in bin/, cnf/make.conf.mac, cnf/make.globals.mac:
- Added preliminary Mac OS X/BSD support.
-
- 06 Jun 2003; Daniel Robbins <drobbins@gentoo.org> ebuild.sh: new eclass
- dependency implementation to totally isolate eclass dependency tweaks
- from in-ebuild dependency tweaks. eclasses now absolutely do not touch
- DEPEND and RDEPEND. Any dependencies added by eclasses are transferred
- to another variable and added in to DEPEND and RDEPEND after the entire
- ebuild has been processed, and after RDEPEND has optionally inherited a
- value from DEPEND due to it being unset. The result of this change is
- that inherit statements no longer need to be placed strategically in an
- ebuild so that they are after the DEPEND and RDEPEND, and fix a host of
- other eclass mis-use/side-effect bugs.
-
-*portage-2.0.48-r1 (29 May 2003): Touchups.
-
- 29 May 2003; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Quick fixes
- to a couple of rare bugs. Added some quoting to $S. Removed the DISTCC
- hosts variable. Moved build-info's creation around so that it actually
- works for non-portage-compiled merges.
-
- 29 May 2003; Nicholas Jones <carpaski@gentoo.org> repoman: Quick fix for
- the files-dir-not-added lack of notification. Added in a fix for a possible
- security problem with the repoman commit messages and symlink-attacks.
-
- 29 May 2003; Nicholas Jones <carpaski@gentoo.org> cnf/*: Added in a diff
- from 'Danny' that contained a SYNC cleanup and expanded explanation. :)
-
- 29 May 2003; Nicholas Jones <carpaski@gentoo.org> output.py: Fixed the
- TitleBar and unset TERM issues.
-
- 29 May 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Added new
- functions to aid in determining virtuals. cpv_all() added to dbapi -- it
- displays all versions instead of just package names -- cp_all() now calls
- this function to gain it's data. get_provide() and get_all_provides() added
- to collect PROVIDE info from vartree packages. getallcpv() added to vartree
- to reference the dbapi call. Always set ignoring errors on findname now.
- Cleaned up the virtuals-trimming code. Don't mention the fixpackages script
- if the user doesn't have any packages.
-
-*portage-2.0.48 (20 May 2003): Cleanups Release.
-
- 20 May 2003; Nicholas Jones <carpaski@gentoo.org> emerge: Tiny cleanups.
- Fixed the profile information in emerge -V for symlinks that have a
- trailing '/' in them. Fix 'emerge -s' filesize lookups to use a new
- function finddigest() from dbapi.
-
- 20 May 2003; Nicholas Jones <carpaski@gentoo.org> emergehelp.py: Updated.
-
- 20 May 2003; Nicholas Jones <carpaski@gentoo.org> etc-update: Small
- fixups for pager issues. Made the nothing-to-do exit message 'happier'.
-
- 20 May 2003; Nicholas Jones <carpaski@gentoo.org> extra-functions.sh:
- Beat the debian-utils requirement out of the functions. Prefers them,
- but falls back to already-existing tools otherwise.
-
- 20 May 2003; Nicholas Jones <carpaski@gentoo.org> fixpackages: 75%
- overhaul to match the new handling of update code.
-
- 20 May 2003; Nicholas Jones <carpaski@gentoo.org> quickpkg: Understands
- "--help" as an option now.
-
- 20 May 2003; Nicholas Jones <carpaski@gentoo.org> man/*: Simple updates.
-
- 20 May 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: fetch()
- now calculates md5sums when doing a fetchonly. finddigest() added for
- finding digests -- gets overlay digests if existing. Binary packages
- are only updated when explicitly requested via fixpackages as a command
- or a FEATURES setting. Clear the dircache after portage initializes to
- reduce memory consumption.
-
-*portage-2.0.48_pre6 (12 Apr 2003): General cleanups and fixes.
-
- 12 May 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Additions
- to better handle overlay ebuilds with regard to metadata cache. aux_get
- calls findname2() which returns (location,in_overlay) -- findname() is
- a wrapper for findname2(). Extra debugging fixes for findname() issues.
-
- 12 May 2003; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Code
- cleanups and a chgrp/chown pass to move all portage-owned files to
- root ownership -- Two pass to leave specific-group and specific-owner
- settings alone.
-
- 12 May 2003; Nicholas Jones <carpaski@gentoo.org> emerge: code cleanups
- and comment touchups.
-
- 12 May 2003; Nicholas Jones <carpaski@gentoo.org> emergehelp.py: Updates
- for --debug, --digest, and --skipfirst.
-
- 12 May 2003; Nicholas Jones <carpaski@gentoo.org> g-cpan.pl: Added more
- arches to the KEYWORDS. Made it copy files back to distfiles.
-
- 12 May 2003; Nicholas Jones <carpaski@gentoo.org> prepstrip: Removed ${D}
- from the outputted filenames during stripping.
-
- 12 May 2003; Nicholas Jones <carpaski@gentoo.org> man/* cnf/*:
- documentation updates.
-
-*portage-2.0.48_pre5 (29 Apr 2003): Cleanups and polish.
-
- 29 Apr 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Another
- typo fix.
-
-*portage-2.0.48_pre4 (28 Apr 2003): Cleanups and polish.
-
- 28 Apr 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Quick fix
- for a traceback/typo in the 'emerge -C' code.
-
-*portage-2.0.48_pre3 (28 Apr 2003): Cleanups and polish.
-
- 28 Apr 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: listdir()
- now takes care of non-existance errors instead of propagating them --- can
- take EmptyOnError=1 as a param to return [] instead of None. All calls to
- portage's listdir() have been updated to the new conventions. listdir()
- ignores .# files from cvs now when ignorecvs=1 is set. Fixes for another
- world-depleting bug. If findname() is passed a virtual, it now informs
- the user to report a bug, instead of weird tracebacks. Added code in
- dblink::unmerge() to remove stale virtual entries when set under trimworld.
-
- 28 Apr 2003; Nicholas Jones <carpaski@gentoo.org> etc-update: enabled the
- diff pager again.
-
- 28 Apr 2003; Nicholas Jones <carpaski@gentoo.org> emerge: --changelog
- implies --pretend. Spelling fixes. Added --skipfirst to allow the first
- package in a resume operation to be skipped over --- allows -e to rebuild
- almost everything, even under weird circumstances. Located another missed
- world-file-depleting bug in unmerge() calls. Included a patch to fix an
- off-by-one bug in depclean.
-
- 27 Apr 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Fix for
- listdir() running on a non-existant dir -- returns None. Sets generated
- files to the portage group -- Manifest & digests. Some coloring additions.
- Handling of 'strict' features added. TEMPORARY addition of 'manifest' USE
- flag so this can go mainstream now. Fixes in doebuild() to get unmerge
- working again for ebuild. Debug code added to findname to help figuring
- out the seemingly random tracebacks -- suspect is pkgsplit(). Unmasking
- code for package.mask added in -- works just like mask, and can override
- specific versions or ranges -- One per line: /etc/portage/package.unmask.
- Security fix for python cPickle code -- mtimedb could create arbitrary
- execution of code bug. Global update notices/info. Fixes for loops over
- listdir() that depended upon raised errors to determine code flow.
-
- 27 Apr 2003; Nicholas Jones <carpaski@gentoo.org> cvstree.py: Added in
- 'removed' status checks.
-
- 27 Apr 2003; Nicholas Jones <carpaski@gentoo.org> repoman: Added in more
- cvs checks. /space/cvsroot is now complained about.
-
- 27 Apr 2003; Nicholas Jones <carpaski@gentoo.org> emerge: Added in
- PORTAGE_CALLER support to identify it as a 'do_update()' candidate.
-
- 27 Apr 2003; Nicholas Jones <carpaski@gentoo.org> dodoc: Check sizes to
- prevent 0 byte files from being added.
-
-*portage-2.0.48_pre2 (10 Apr 2003): Digests & Repoman
-
- 16 Apr 2003; Nicholas Jones <carpaski@gentoo.org> repoman: Added CVS/Root
- checks to ensure Manifests will be correct. Fixed local use-flag bug.
- Commented out the no-stable code, as it can't be implemented well as is.
-
- 16 Apr 2003; Nicholas Jones <carpaski@gentoo.org> prepstrip: Fixed a
- problem where files were not being stripped.
-
- 16 Apr 2003; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Added in
- a call to 'make -n' when under the --debug flag.
-
-*portage-2.0.48_pre1 (10 Apr 2003): Digests & Repoman
-
- 10 Apr 2003; Nicholas Jones <carpaski@gentoo.org> prepstrip: fix it so
- that it actually strips and provides reasonable info.
-
- 10 Apr 2003; Nicholas Jones <carpaski@gentoo.org> repoman: Added short
- option for pretend. Added pretend capability to the majority of the
- commit code. Bug fix for local IUSE code. Added in support for removed
- cvs files.
-
- 10 Apr 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Changed
- 'manifest' to 'Manifest'. Removed the requirement that Manifests exist --
- complain about it, but only die if 'strict' is set.
-
-*portage-2.0.47-r15 (09 Apr 2003): New digests/manifests, touchups.
-
- 09 Apr 2003; Nicholas Jones <carpaski@gentoo.org> bin/*: Fixes for Cross-
- compiling.
-
- 09 Apr 2003; Nicholas Jones <carpaski@gentoo.org> repoman: Quick touchups
- in the category- and repository-level manifest calls.
-
- 09 Apr 2003; Nicholas Jones <carpaski@gentoo.org> cnf/*: Added/sync'd
- PORTAGE_NICENESS.
-
- 09 Apr 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Fixed hole
- in the listdir code that occured on cache expiration in the -r14 changes.
-
-*portage-2.0.47-r14 (09 Apr 2003):
-
- 09 Apr 2003; Nicholas Jones <carpaski@gentoo.org> bin/*: Removed DEBUGBUILD
- in favor of FEATURES/RESTRICT settings of nostrip.
-
- 09 Apr 2003; Nicholas Jones <carpaski@gentoo.org> ebuild: Removed root
- restriction so that digests can be made as non-root.
-
- 09 Apr 2003; Nicholas Jones <carpaski@gentoo.org> emerge: Added niceness
-
- 09 Apr 2003; Nicholas Jones <carpaski@gentoo.org> repoman: Added the
- nostable/allmasked check in. Small cleanups. Move to 'manifests' instead
- of enhanced digests. Added filters on $Id and $Id. Fixed commitmsg
- and added short options.
-
- 09 Apr 2003; Nicholas Jones <carpaski@gentoo.org> cvstree.py: More features
- added. pathdata() provides dir or file info. isadded() gives cvs status.
-
- 09 Apr 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: listdir()
- enhancements. Debug removal. Broke up the digest creation functions to
- make them a little more generic. Added manifest code into the digest
- calls. Moved code in doebuild() around to allow for non-root calls for
- devs and repoman. Added 'PORTAGE_CALLER' env var to prevent repeated
- running of unnecessary functions.
-
-*portage-2.0.47-r13 (02 Apr 2003): Fixes & Security -- Repoman+Digests
-
- 02 Apr 2003; Nicholas Jones <carpaski@gentoo.org> emerge: Added debug to
- the dep selection code in depgraph.
-
- 02 Apr 2003; Nicholas Jones <carpaski@gentoo.org> cvstree.py: Added new
- module to handle cvs information without having to ask 'cvs status' many
- times to determine file locations.
-
- 02 Apr 2003; Nicholas Jones <carpaski@gentoo.org> repoman: Major fix ups
- to ease the introduction of enhanced digests. Intelligent cvs queries
- and use of the new cvstree module for information about files in the tree.
- Auto-digests, auto-fetches trivial cvs changes, detects changelogs that
- exist but have not been added to cvs, and detects '*' in KEYWORDS.
-
- 02 Apr 2003; Nicholas Jones <carpaski@gentoo.org> *: A typo fix or two.
- prepstrip: fixup for sed's issues with '\000' (made it \001).
-
-*portage-2.0.47-r12 (27 Mar 2003): Fixes & Security -- Repoman+Digests
-
- 27 Mar 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Moved all
- initialization error messages to stderr.write() calls so that the output
- doesn't get mixed will called data. Added a warning about not being in
- the portage group. Color touchups. Digest path fixups.
-
- 27 Mar 2003; Nicholas Jones <carpaski@gentoo.org> repoman: Significant
- additions to the commit code to nicely handle the changes for enhanced
- digests. Added --commitmsg and --commitmsgfile flags to allow easier
- scripting. Added a REPOMAN environment variable to prevent portage from
- running do_updates(). Added 'grouplist' which makes groupings of subparts
- from a list of parts. Added do* functions to manage the recursion into
- the dirs of the repository.
-
- 27 Mar 2003; Nicholas Jones <carpaski@gentoo.org> xpak.py: Change from
- lstat to stat... We're not concerned with links. Return 0 on not found.
-
- 27 Mar 2003; Nicholas Jones <carpaski@gentoo.org> portageq: added new
- functions: best_visible, mass_best_visible, all_best_visible.
-
- 27 Mar 2003; Nicholas Jones <carpaski@gentoo.org> emerge: Added '--digest'
- to force recreation of digests from the command line. Fixed portage version
- comparisions for the 'update portage please' notices.
-
- 27 Mar 2003; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Added an
- extra rm of the 'successful' file.
-
- 26 Mar 2003; Alain Penders <alain@gentoo.org> portageq: added
- mass_best_version method to help GUIs resolve best versions.
-
-*portage-2.0.47-r11 (22 Mar 2003): Fixes & Security -- Winding up 2.0 series.
-
- 22 Mar 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: listdir()
- now is capable of recursive calls and listing only files. Used in new
- additions to digestgen() and digestcheck() for ebuilds and files/* --
- listdir(dir,recursive=0,filesonly=0). Digests now must contain all files
- from files/ and also the ebuild -- non-archives are prefixed with '/'.
- Full tbz2 fixes are now implemented... They can take a while to perform
- so status thingies are provided. update_ents() provides a batch-update
- as it would take rediculous amounts of time otherwise.
-
- 22 Mar 2003; Nicholas Jones <carpaski@gentoo.org> : fixpackages calls
- portageexit() to save the mtimedb state so it actually works. :)
-
- 22 Mar 2003; Nicholas Jones <carpaski@gentoo.org> emerge: Added
- 'xtermTitle' calls to emergelog to set the title during phases of merging.
- Added more emergelog calls to provide more specific information.
-
- 22 Mar 2003; Nicholas Jones <carpaski@gentoo.org> : Added two scripts to
- aid portage. 'fixdbentries' takes (old, new, path) and does an inteligent
- sed on all the db entries that should be changed. Used by global moves and
- in the tbz2 fixes. 'fixpackages' expires the updates timestamp and reloads
- portage to force do_updates() to run. Added missing Header lines.
-
- 22 Mar 2003; Nicholas Jones <carpaski@gentoo.org> xpak.py: Correct off-by-8
- bug in last commit. infosize does not contain the full size... Offset
- begins _after_ the marker -- Added xpaksize for complete offset from EOF.
-
- 22 Mar 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Fix for
- die-on-wheel-group-missing bug. The logfiles don't cause the counter
- to be incremented and now match the merged package's COUNTER value.
- get_counter_tick_core(root) does not increment the counter. Sandbox is
- now working in all enabled cases, instead of just usersandbox. /var/db
- and binaries (assuming they are RW) are now updated to contain the proper
- info --- Added a 'fixdbentries' script to perform the updates to the data,
- and added a 'move_ent' function to class binarytree (not fakedbapi). Added
- a 'ebuild does not exist' error for the case of ebuilds in a wrongly named
- directory. Added a notice for invalid tbz2's. Removed the 'make.defaults'
- missing notice :-/. Duplicated the bintree creation so that it would be
- available for do_update(). Catch when 'updates' is missing from mtimedb.
-
- 22 Mar 2003; Nicholas Jones <carpaski@gentoo.org> output.py: Added xterm
- title bar function to set the titles... xtermTitle(mystring). Only set
- if using color and terminal is [axE]term.
-
- 22 Mar 2003; Nicholas Jones <carpaski@gentoo.org> repoman: Additions
- courtesy of Gerk and Vladamir... Checks all portage variables. Runs
- a check against invalid (but still parsable) syntax errors in names.
- Checks for invalid IUSE from use.desc and use.local.desc. Checks that
- licenses are valid. Checks for legal keywords with a default set or
- keywords.desc, if available. 'missingvar' tests moved to aux_get() calls.
-
- 22 Mar 2003; Nicholas Jones <carpaski@gentoo.org> prepstrip: Added test
- to allow cross-compile-stripping to work properly. Rewrite the script to
- be whitespace-aware and recursively process directories at level 1.
-
- 22 Mar 2003; Nicholas Jones <carpaski@gentoo.org> prepallstrip: Removed
- all code from this script and made it a call to "prepstrip $D".
-
- 22 Mar 2003; Nicholas Jones <carpaski@gentoo.org> g-cpan.pl: Add in chomps
- for the portageq calls.
-
- 22 Mar 2003; Nicholas Jones <carpaski@gentoo.org> emerge: Allow
- --fetchonly of block'd packages.
-
- 22 Mar 2003; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Cleanups and
- corrections for sandbox which has been disabled and broken for a couple
- revisions now. use_{enable,with} die messages removed as they are in sub-
- shells when used and won't kill the merge --- Moved to echo >&2.
-
- 21 Mar 2003; Nicholas Jones <carpaski@gentoo.org> xpak.py: Updates to make
- the xpak code self-sustaining. Added basic documentation to the functions
- and a description of the format in the comments at the top of the module.
-
-*portage-2.0.47-r10 (13 Mar 2003): Fixes -- Winding up 2.0 series.
-
- 13 Mar 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Less strict
- check of permissions when fixing permissions for userpriv -- only requires
- 02070 and group portage instead of 02770 for check.
-
- 13 Mar 2003; Nicholas Jones <carpaski@gentoo.org> g-cpan.pl: Fix to the
- call to portageq.
-
- 13 Mar 2003; Nicholas Jones <carpaski@gentoo.org> etc-update: Fix for
- deleted files in a CONFIG_PROTECT situation.
-
- 13 Mar 2003; Nicholas Jones <carpaski@gentoo.org> emerge: fix for keepwork
- and keeptemp ** MUST ENSURE THAT CLEAN CLEANS BEFORE A FULL MERGE.
-
- 12 Mar 2003; Nicholas Jones <carpaski@gentoo.org> emerge: Added more
- conditions on rsync -- Proper error detection. Rsync failures will not
- cause a cache regeneration. Portage no longer considers masked portages
- candidates for 'An update to portage'.
-
-*portage-2.0.47-r9 (10 Mar 2003): Fixes -- cvs-src, ebuild.sh, repoman
-
- 10 Mar 2003; Nicholas Jones <carpaski@gentoo.org> make.conf*: Adjusted
- the LOGDIR message and change the default location to /var/log/portage.
-
- 10 Mar 2003; Nicholas Jones <carpaski@gentoo.org> repoman: One more fix
- for the PORTDIR setting -- '/usr/portage' was matching '/usr/portage.cvs'.
-
- 10 Mar 2003; Nicholas Jones <carpaski@gentoo.org> etc-update: Added the
- capability to automerge all files in the list -- with out without prompts.
-
- 10 Mar 2003; Nicholas Jones <carpaski@gentoo.org> emerge: Added messages
- on failure to create a depgraph -- Helps with bad DEPEND detection. Made
- 'emerge sync' stop if it fails, and not update cache.
-
- 10 Mar 2003; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: added an
- external extra_functions.sh file with extra functions for portage --
- it is flag-o-matic and pieces of eutils presently. Removed libdir/incdir
- from einstall(). Added 'keepwork' to keep the source code after a package
- is merged. Removed the 'local' declaration of ROOT in dyn_install() --
- this fixes has_version in src_install(). Fixed the sourcing of environment
- so that variables are peristent across stages of a merge. Removed the
- SANDBOX vars from the environment file, as that upsets SANDBOX.
-
- 10 Mar 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Fixes to
- permission problems with cvs-src and ccache (removed nested try's). Log
- counter fix -- One log per package now.
-
-*portage-2.0.47-r8 (02 Mar 2003): Fixes -- cvs-src, ebuild.sh, repoman
-
- 02 Mar 2003; Nicholas Jones <carpaski@gentoo.org> *: Fixes to eclass
- errors and messages -- raise on not found. emerge --debug works for
- ebuilds with syntax errors now -- needs better implimentation though,
- debug=0 in params can't be used due to scope limits.
-
- 28 Feb 2003; Nicholas Jones <carpaski@gentoo.org> *: reorg ebuild.sh again
- to ensure that aliases are expanded properly. Repoman now determines the
- proper PORTDIR to set when loading portage to scan the cvs tree. portage.py
- fixes for cvs-src permission problems with userpriv.
-
-*portage-2.0.47-r7 (27 Feb 2003): Quick fix for eerror() problem.
-
-*portage-2.0.47-r6 (26 Feb 2003): Drop deprecated warnings for new syntax.
-
- 26 Feb 2003; Alain Penders <alain@gentoo.org> emerge: Drop deprecated
- warnings for new syntax. Don't make people switch to the new syntax
- until it's agreed on by everyone.
-
-*portage-2.0.47-r5 (26 Feb 2003): Wicked Fast, BugFree v3.1, UserPriv
-
- 26 Feb 2003; Nicholas Jones <carpaski@gentoo.org> *: Update docs and help
- again to drop --system and --world, and pull them from emerge's command
- line. General cleanups.
-
- 26 Feb 2003; Nicholas Jones <carpaski@gentoo.org> emerge/portage.py: Fixes
- to the cache directory permissions -- all dirs are properly owned by group
- portage now. Add CCACHE_SIZE setting so we set a value for ccache. If the
- dirs are set to the wrong perms, make sure they get changed recursively.
-
- 26 Feb 2003; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Rearraged the
- code blocks so that they are easier to follow -- no more code in between
- functions. All code follows function definitions now.
-
-*portage-2.0.47-r4 (25 Feb 2003): Wicked Fast, BugFree v3, UserPriv
-
- 25 Feb 2003; Nicholas Jones <carpaski@gentoo.org> *: Update docs and help
- to match the deprecation of actions without '--' preceding them.
-
- 25 Feb 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Message
- updates. Fix secpass for portage user. Set PORTAGE_GID for ebuild.sh to
- use regardless of GID in use (wheel/portage). Only try and delete things
- in PORTAGE_TMPDIR if we have perms to do it... see secpass note. Moved the
- cachedir creation after settings in created so that we can use spawn instead
- of system for the calls to 'chown/chmod -R'. Fix the note on make.defaults.
- Fix for the 404 catcher in the fetch code.
-
- 25 Feb 2003; Nicholas Jones <carpaski@gentoo.org> emerge: Move all actions
- into '--action' form -- Notify of deprecated usages. Moved the functions
- from the top of emerge down below the command line parser for readability.
- Kill FEATURES=noauto if we're running emerge -- it can break things. Make
- -U imply -u so people aren't confused. --debug now enables all debug
- variables in portage. Fix wheel/portage group requirements messages. Get
- the current portage version when doing rsync via the portage tree's files
- and not the loaded cache so that it is aware of changes immediately instead
- of the next run.
-
- 25 Feb 2003; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Try to change
- to good a good path at the start of the ebuild to prevent operations from
- happening in places like home. After the ebuild's install phase, check to
- see if any unsafe perm combo's exist and die if they do. Fix permissions
- on the cachedir as we're changing them to root:portage. &> on environment
- file prevented it from actually creating the file -- changed to 2>. Change
- 'true' to 'exit 0' to be a bit more explicit about what we're doing.
-
- 24 Feb 2003; Alain Penders <alain@gentoo.org> repoman: Added nested/sub-shell
- die testing.
-
- 23 Feb 2003; Alain Penders <alain@gentoo.org>: Added bin/portageq tool to
- provide access to portage internal information without using APIs that
- are changing. All tools that currently access portage information by
- importing portage.py and poking around in it should switch to using this
- tool instead. Changed some tools in bin/ to use it, so those won't break
- either.
-
-*portage-2.0.47-r3 (17 Feb 2003): Wicked Fast, BugFree v2, UserPriv
-
- 22 Feb 2003; Nicholas Jones <carpaski@gentoo.org> bin/: added dispatch-conf
- and db-fix.pl to the bins. Looking at dispatch-conf to replace etc-update
- written by Jeremy Wohl (bug 14079). db-fix.pl is a rescue tool that Blizzy
- wrote up to recreate/repair counter files in the DB -- shouldn't be needed
- anymore as portage fixes the counter at every load of the counter, but we
- will hang on to it just in case.
-
- 22 Feb 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: moved the
- wheelgid references to portage_gid references. Wheel is being deprecated.
- Added notice to baselayout message that group portage controls everything
- now. Killed the BASH_ENV settings -- moved the reference to /etc/portage/
- in case it's desired in some way -- This fixes AROUND ONE HUNDRED reports
- on 2.0.47-r2. Try/catch invalid tbz2's -- should fix a couple bad tarball
- bugs. Make the prelink call spawn'd without sandbox. buildpkg/userpriv
- fix via actionmap (Jasmin Buchert; bug 16106).
-
- 22 Feb 2003; Nicholas Jones <carpaski@gentoo.org> emerge: --upgradeonly
- added from jrray's patch; added some slot handling. --deep got a short
- flag 'D'. 'info' can be run by non-root. '--quiet' kills all but package
- names on searches. Basic slot detection added to the pretend output so
- that it doesn't always say downgrade. Ignore block list when fetching.
- Can use RSYNC_EXCLUDEFROM to select a from for rsync to --exclude-from;
- (Michael Sterrett; Bug 15882). 'emerge rsync' deletes the dbcachedir
- before it begins updates. Catch exceptions during rsync cache regen.
-
- 22 Feb 2003; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: General fixes
- to the spelling of things. 'cd $PORT_TMPDIR' corrected to the actual vars
- name. Stripping distcc from PATH if it is there before portage sets it.
- Same with ccache. If we try to compile with out unpacking, complain. Make
- sure to chown files to user portage.
-
- 22 Feb 2003; Nicholas Jones <carpaski@gentoo.org> cnf/*: CFLAGS notes,
- pentium4 breaks things. Added more descriptions for FEATURES. Added
- RSYNC_EXCLUDEFROM description.
-
- 22 Feb 2003; Nicholas Jones <carpaski@gentoo.org> *: Updates to the
- copyright headers.
-
-*portage-2.0.47-r2 (17 Feb 2003): Wicked Fast, BugFree, UserPriv
-
- 18 Feb 2003; Alain Penders <alain@gentoo.org> portage-2.0.47-r2.ebuild:
- If $PORTAGE_TEST is set, skip the beeps/delay messages. Needed to make
- the regression test scripts bearable :-)
-
- 17 Feb 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Fixed the
- depcache code once again -- ebuild and cache were never set to the same
- mtimes. Removed last os.system() call.
-
- 17 Feb 2003; Nicholas Jones <carpaski@gentoo.org> repoman: catdir fix.
- Added 'emerge-webrsync' from gentoolkit so users can use snapshots on
- initial installs.
-
- 17 Feb 2003; Nicholas Jones <carpaski@gentoo.org> emergehelp.py: Updated
- it to the current calls and conventions. Added comments about etc-update
- and emerge-webrsync.
-
- 17 Feb 2003; Nicholas Jones <carpaski@gentoo.org> emerge: Changed the
- search code around to use compiled re's instead of .lower() searches.
- Trivial touchups in spacing. Only write worldfile if not --pretend. Fix
- losing-packages-to-be-merged during merge bug. Allow options to change
- in a --resume call so that --buildpkg and such may be added.
-
- 17 Feb 2003; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Added a
- PREROOTPATH to allow special paths to be added prior to root and ebuild's
- required pathes. ${S} is now generated prior to sourcing an ebuild so
- that it is available in the ebuild at source-time, as opposed to in the
- phases.
-
-*portage-2.0.47-r1 (16 Feb 2003): Wicked Fast + UserPriv
-
- 16 Feb 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Changed a
- few os.system calls to spawn calls. Added userpriv in features check to
- spawn. HOME is now placed in BUILD_PREFIX/homedir when userpriv is enabled.
- Valid command for doebuild() moved to the top of the function. Reorganized
- the directory creation code in doebuild(). HOME is cleaned for every merge
- if in userpriv. LOG_COUNTER fix for the per-ebuild logs -- Only one value
- is used per $PF now.
-
- 16 Feb 2003; Nicholas Jones <carpaski@gentoo.org> emerge: Adjusted options
- to remove --autoclean and make it on-by-default unless killed in FEATURES
- with 'noclean'. Removed debug for --resume. Added RSYNC_TIMEOUT as some
- dialup and other users are experiencing problems with rsync never finishing
- a connection. Added check at the end of an rsync to notify the user when
- a new version of portage is available. Added --pretend capability/fix to
- the --resume code. No resume data and a --resume is no longer a failure.
-
- 16 Feb 2003; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Updates to
- logic of ${T}/successful handling to allow proper cleaning. dyn_clean()
- will do a recursive dir removal after cleaning specific files. Recursive
- chown removed. CCACHE_DIR handled in portage.py now. Removed a few comments
- and unnecessary checks.
-
- 16 Feb 2003; Nicholas Jones <carpaski@gentoo.org> cnf/*: added comments
- on FEATURES=noclean,noauto,userpriv,usersandbox
-
- 15 Feb 2003; Nicholas Jones <carpaski@gentoo.org> emerge: Alain changed
- the .config() call in the search class to a .settings[] -- Speedup.
-
- 14 Feb 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Added
- resume to mtimedbkeys.
-
- 14 Feb 2003; Nicholas Jones <carpaski@gentoo.org> emerge: --columns
- flag to support column output in --pretend. --resume support added
- to restart portage with the package list with which it stopped -- data
- is stored in mtimedb. Added '--verbose' to 'emerge info' to display
- all variables in settings.
-
- 13 Feb 2003; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Fix to
- distcc and ccache to prevent them from calling themselves if they
- catch themselves in the path -- double path problem.
-
- 13 Feb 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Fixes to
- the aux_get() code for metacache updates to eclass deps. Debug output
- removal.
-
- 13 Feb 2003; Nicholas Jones <carpaski@gentoo.org> emerge: More fixes
- to the calls to aux_get() for metacache code. Rearranged the option and
- action arrays. Added --nospinner flag.
-
- 13 Feb 2003; Nicholas Jones <carpaski@gentoo.org> emerge: Reorg of flags.
- Addition of --noconfmem to prevent portage from yanking already-merged
- config files. Extra message in --version about missing gcc and sourcing
- /etc/profile to get it. Yanked the majority of the rsync cachedb update
- code -- Moved it to aux_get.
-
- 13 Feb 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Typo fixes.
- aux_get() now has a metacachedb param for using metadata cache over the
- initial call to doebuild() -- Should speed up rsync users regen time.
- Added support for --noconfmem via settings["NOCONFMEM"] in treewalk().
- Added sys.exit(1) on failure to move files during merge. Comment
- realignment.
-
- 11 Feb 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Fix the
- depcache code by adding in a missing stat call after regeneration.
- Call portageexit() in exithandler() so that mtimedb gets written out
- on cancelled runs. Add in --quiet option. Currently only kills the
- processing messages on emerge regen.
-
- 11 Feb 2003; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Fix the
- eclass code to remove the double inherits cause by eclasses managing
- that on their own.
-
-*portage-2.0.46-r12 (07 Feb 2003): Fixups
-
- 07 Feb 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Move
- logdir check/generation code so that it's created before 'clean'. Fix
- the usemask/archkeys bug and a related bug with the var enabled. Fix
- to movefile() code to ensure symlinks are handled properly.
-
- 07 Feb 2003; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Make per-
- ebuild logs use the counter to get a chronological list instead of a
- timestamp based one.
-
- 07 Feb 2003; Nicholas Jones <carpaski@gentoo.org> emerge: If a package
- is requested to be updated but is not merged, still add it to world.
-
- 07 Feb 2003; Nicholas Jones <carpaski@gentoo.org> *: Update to help.
- Mostly comments on --verbose. Clarification on make.conf* mirrorselect.
-
-*portage-2.0.46-r11 (04 Feb 2003): Fixups
-
- 04 Feb 2003; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: distcc
- fixups along with ccache fixups from the bug 13897 effort. Added more
- varaibles to the build data output. Allow CC/CXX to be set if they are
- not set in environment.
-
- 04 Feb 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Removed an
- unnecessary call to os.unlink() in movefile that resulted in it failing
- when moving files across devices. HTML 404 catcher -- if the distfile is
- a reasonable size, and it's got <title>.*(not found|404).*</title> in it
- it will be deleted and the next mirror persued.
-
-*portage-2.0.46-r10 (03 Feb 2003): Feature/Function Cleanups
-
- 03 Feb 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Added
- quotes to fallback checksum code to ensure it works on space-containing
- filenames. Error message added in writedict. Worldfile-depleeting bug
- fixed -- 'emerge -e world'+AUTOCLEAN would cause every package in world
- to be removed from the worldfile to be removed.
-
- 03 Feb 2003; Nicholas Jones <carpaski@gentoo.org> make.conf: updated the
- commentary about AUTOCLEAN.
-
- 03 Feb 2003; Nicholas Jones <carpaski@gentoo.org> etc-update: Added in
- automerge functionality for trivial changes. Comments and whitespace
- are just merged without asking when it's enabled.
-
- 03 Feb 2003; Nicholas Jones <carpaski@gentoo.org> emerge: redundant cmd
- line flags warning. Show enabled/disabled USE flags on -vp (Masatomo
- Nakano). Missing changelog fix for --changelog. Added an rsync timeout
- of 60 seconds and a message on fail due to timeout.
-
- 03 Feb 2003; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: CC/CXX fix
- to override profile.env settings. FEATURES=distcc support to get past
- profile.env's CC settings/lockdown. If CBUILD is defined, it is appended
- to econf's output in a --build= statement. Added .unpacked marker to
- the unpack code to ensure a proper unpack phase.
-
- 01 Feb 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Adjusted
- the arch.list code to be a general masking setup. /etc/make.profile/
- and /etc/portage/ have use.mask files that are concatenated. On fetch,
- if downloader reports failure, then check that filesize<digestfilesize
- and continue onto the next mirror before dying.
-
- 27 Jan 2003; Nicholas Jones <carpaski@gentoo.org> ebuild.sh/portage.py:
- eclass fixups for newdepend issues where newdepend operates on RDEPEND
- prior to it being set to default by ebuild.sh. Repoman fix for multi-
- arch failures by excluding /usr/portage/profiles/arch.list entries from
- consideration, excluding $ARCH.
-
-*portage-2.0.46-r9 (14 Jan 2003): portage restart fix
- 15 Jan 2003; Nicholas Jones <carpaski@gentoo.org> emerge: fixed the
- missing tabs from the cvs diff backport.
-
-*portage-2.0.46-r8 (14 Jan 2003): touchup KV{,ERS} for public stable
-
-*portage-2.0.46-r7 (14 Jan 2003): Backport from cvs and fixes.
- portage-2.0.46* :: EXCUDES ALL PORTIONS RELATING SPECIFICALLY TO USERPRIV
-
- 14 Jan 2003; Nicholas Jones <carpaski@gentoo.org> portage.py: Persistant
- KV and KVERS on depend calls. Sandbox violation of do_upgrade() caught.
-
- 14 Jan 2003; Nicholas Jones <carpaski@gentoo.org> emerge: --buildpkg
- touchups and forward porting of some of 2.0.46-r6. Restart on -r0/proper
- versions of portage due to VERSION not matching pkgsplit output fixed.
-
- 14 Jan 2003; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: 20% speedup
- in dep generation by removing all external app calls. Moved KVERS to
- portage to make it persistent during dep calls. Other code moved into
- conditionals based on $*!=depend. chmod's swaped with umask.
-
- 11 Jan 2003; Jack Morgan <jmorgan@gentoo.org> cnf/make.conf.sparc:
- updated make.conf.sparc to add CFLAGS for sparc32 and sparc64
-
- 08 Jan 2003; Nicholas Jones <carpaski@gentoo.org> bin/g-cpan.pl:
- Added from bug 3450 -- Creates and merges perl module ebuilds on-the-fly
- from cpan and merges them.
-
- 08 Jan 2003; Mark Guertin <gerk@gentoo.org> cnf/make.conf.ppc:
- updated make.conf.ppc to remove G3 options for CFLAGS and other
- small tweaks
-
-*portage-2.0.47 (06 Jan 2003): UserPriv
-
- 06 Jan 2003; Nicholas Jones <carpaski@gentoo.org> portage.py:
- Forced HOME to BUILD_PREFIX regardless of user. Added in a 'rm -Rf' notice
- to readonly-fs/Full-Disk notices. Disabled the usepkg on buildpkg function
- of portage. digraph.hasallzeros() added -- Determines if tree is zero depth.
-
- 06 Jan 2003; Nicholas Jones <carpaski@gentoo.org> cnf/*: Added in the
- PORT_LOGDIR option and description. Also added a GENTOO_MIRRORS section
- back into the make.conf* files so that users are aware that they need to
- specify ibiblio. Killed the linefeeds in the sparc conf.
-
- 06 Jan 2003; Nicholas Jones <carpaski@gentoo.org> emerge: options and
- actions fixup. Added new short options and long options. --buildpkgonly
- now works and restricts the process to deplists of zero depth. Updated
- the help for portage, and created a seperate shorthelp function with
- the terse, options-only, versions of the regular help. Fixed the env_update
- call after --fetchonly runs.
-
- 06 Jan 2003; Nicholas Jones <carpaski@gentoo.org> ebuild.sh:
- touchups to the user priv and logging code to kill the invalid user
- messages when portage:portage doesn't exist on the system.
-
-*portage-2.0.47_pre4 (03 Jan 2003): UserPriv Works + Per-Ebuild logging
-
- 04 Jan 2003; Nicholas Jones <carpaski@gentoo.org> portage.py:
- tokenize() fixups to remedy the || reduce problems.
-
- 04 Jan 2003; Nicholas Jones <carpaski@gentoo.org> prepall:
- Fixup for '//' problem.
-
- 03 Jan 2003; Nicholas Jones <carpaski@gentoo.org> ebuild.sh:
- Support for per-ebuild logging via PORT_LOGDIR variable. ebuild.sh
- now can do usermode compiles _and_ maintain the enviroment with a
- few minor restrictions like changing portage variables.
-
-*portage-2.0.47_pre3 (02 Jan 2003): Forward port of fixes & New fixes
-
- 02 Jan 2003; Nicholas Jones <carpaski@gentoo.org> emerge:
- tbz2 handling improved -- checks current, pkgdir/All/x, pkgdir/x.
- -r0 removal on current ver to stop people from complaining about it.
- More info added to 'emerge info' output.
-
- 02 Jan 2003; Nicholas Jones <carpaski@gentoo.org> portage.py:
- Spaces in filenames fix for prelink-capable systems. Exception caused
- by invalid or unsatisfiable '||' dependancies caught and warned about.
- auxdbkey order fix. dbcachedir now ensures a '/' before the cache dir.
-
- 02 Jan 2003; Nicholas Jones <carpaski@gentoo.org> prep*:
- Fixes from Azarah to make the included files more proper.
-
- 26 Dec 2002; Phil Bordelon <sunflare@gentoo.org> man/emerge.1:
- Added documentation of the --deep option.
-
-*portage-2.0.46-r5 (30 Dec 2002): Touchups and sparc confs
-
- 30 Dec 2002; Nicholas Jones <carpaski@gentoo.org> *:
- fix for prelink unmerge problems. small touches to outputs.
- sparc configs added. Azarah's fixups for the prep* scripts.
- Changes to 'emerge info' output. tbz2 prefixes PKGDIR/All if the
- tbz2 doesn't exist in the current dir.
-
-*portage-2.0.46-r3 (24 Dec 2002): Backport of fixes in 2.0.47_pre2
-
-*portage-2.0.47_pre2 (24 Dec 2002): Feature: userpriv compiles
-
- Tokenizer fixup courtasy of Evgeny Roubinchtein. unalias -a in ebuild.sh.
- Moved the help() to emergehelp.py. prepallman fixup for missed symlinks.
- More info provided for portage-user-missing message. FEATURES=sandboxuser
- not provides sandbox in compile phase along with userpriv. mtimedb
- exception should actually be caught now.
-
-*portage-2.0.47_pre1 (21 Dec 2002): Feature: userpriv compiles
-
- 21 Dec 2002; Nicholas Jones <carpaski@gentoo.org> *:
- ebuild.sh: portage user setup. dyn_setup is always run before calls to
- unpack,compile,install. emerge: infodirs cleanup, traceback on invalid
- dir fix. Proper regen on info change. portage.py: uid/gid discovery and
- check code. PRELINK_PATH and PRELINK_PATH_MASK added and renamed. Made
- HOME set to BUILD_PREFIX when HOME is unset. Patched in UserPriv compile
- code and checks and complaints to ensure smooth integration. Reorganized
- chown calls to ensure things get set right for userpriv and not. Sandbox:
- UID check to prevent non-root errors patched in.
-
-*portage-2.0.46-r2 (18 Dec 2002): Feature Stable Release Prelink + Bug Fixes
-
- 18 Dec 2002; Nicholas Jones <carpaski@gentoo.org> portage.py, emerge:
- Traceback fix for blocking packages in the restart check in emerge. Added
- missing import for commands in portage.py.
-
-*portage-2.0.46 (18 Dec 2002): Feature Stable Release Prelink + Bug Fixes
-
- 18 Dec 2002; Nicholas Jones <carpaski@gentoo.org> portage.py:
- PRELINK_MASK adds ignore paths for prelink.conf. Error message
- adjustments for appearance/readability. auxdbkey changes.
-
- 18 Dec 2002; Nicholas Jones <carpaski@gentoo.org> emerge:
- --changelog displays changelogs of packages on update. gcc version
- fix for --version. Missing indent in masked output corrected. Some
- color additions. Prelink code removed -- Users can do it instead.
-
- 18 Dec 2002; Nicholas Jones <carpaski@gentoo.org> ebuild.sh:
- use_{enable,with}() work now. Rearranged auxdbkeys again.
-
- 17 Dec 2002; Martin Schlemmer <azarah@gentoo.org> bin/ebuild.sh:
- Regenerate /lib/cpp and /usr/bin/cc in pkg_setup if they are not
- files to ease the broken pkg_postrm() some gcc have.
-
-*portage-2.0.46_pre2 (15 Dec 2002): Feature Prerelease: prelink
-
- 15 Dec 2002; Nicholas Jones <carpaski@gentoo.org> emerge/portage.py:
- Fixups. PRELINK_MASK code. 'prelink -af' in post_emerge().
-
- 15 Dec 2002; Mark Guertin <gerk@gentoo.org> cnf.make.conf.ppc:
- updated incorrect CFLAG option and appended -mabi=altivec info
-
-*portage-2.0.46_pre1 (15 Dec 2002): Feature Prerelease: prelink
-
- 15 Dec 2002; Nicholas Jones <carpaski@gentoo.org> portage.py: Counter
- fixups. Prelink code.
-
- 15 Dec 2002; Nicholas Jones <carpaski@gentoo.org> bin/f*: Made them
- loop over multiple files instead of just doing one.
-
- 13 Dec 2002; Nicholas Jones <carpaski@gentoo.org> emerge/portage.py:
- Added disabled prelink code. Needs a little more work. Gave portage the
- ability to restart on upgrades to the portage version. Ebuilds for prior
- versions will intentionally die on upgrade. USE=build disables it so
- bootstrap isn't affected.
-
-*portage-2.0.45-r5 (13 Dec 2002): Feature addition: PDEPEND+fixes
-
- 13 Dec 2002; Nicholas Jones <carpaski@gentoo.org> emerge/portage.py:
- Added restart on portage upgrade code and adjusted ebuild to accomplish
- this until version matches -r5.
-
-*portage-2.0.45-r4 (11 Dec 2002): Feature addition: PDEPEND
-
- 11 Dec 2002; Nicholas Jones <carpaski@gentoo.org> ebuild.sh/portage.py:
- Added [CDEPEND, PDEPEND, REBUILD] to auxdbkeys and updated ebuild.sh.
- Prevented autoclean on fetchonly.
-
- 11 Dec 2002; Nicholas Jones <carpaski@gentoo.org> emerge:
- Added support for PDEPEND. Rewrote some dep handling variables for
- readability and ease of use.
-
-*portage-2.0.45-r3 (09 Dec 2002): Touchups.
-
- 09 Dec 2002; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Fixed
- use_enable() and use_with() so that they actually work -- also don't
- have to specify 2nd parameter, 1st is assumed.
-
- 09 Dec 2002; Nicholas Jones <carpaski@gentoo.org> emerge: Added previous
- version/downgrade messages to --pretend. Extra message to notify of dep
- calculation failure as bad deps can exit portage without error.
-
- 09 Dec 2002; Nicholas Jones <carpaski@gentoo.org> portage.py: Attempt
- at fixing the sandbox-crashes-missing-HOME bug. Added a try/catch around
- the virts/'del x' code to prevent tracebacks on boxes without a virtuals
- file.
-
- 06 Dec 2002; Nicholas Jones <carpaski@gentoo.org> emerge: Fixed 'xfrom'
- variable-used-before-assignment message on masked ebuilds.
-
- 06 Dec 2002; Nicholas Jones <carpaski@gentoo.org> portage.py: Disabled
- writes to DBs when sandbox is enabled.
-
- 03 Dec 2002; Nicholas Jones <carpaski@gentoo.org> portage.py: Fixed a
- nice bug where python was writing 'L' at the end of long() types for
- mtimes.
-
-*portage-2.0.45 (02 Dec 2002): More bug fixes for stable.
-
- 02 Dec 2002; Nicholas Jones <carpaski@gentoo.org> portage.py, output.py:
- Added several color options and functions to output.py. 'DO NOT EDIT'
- notices added to .env generated profile settings files. Touchups and
- nofetch() call for restricted fetches. Moved some file IO code for
- SLOTs into a try/catch. { mtimedb fixups. Eclass aux_get() error fix.
- aux_get() error fix via try/catch with magically-missing ebuild when
- doing stat on it. Enabled full eclass()/inherit code. Added code to
- flushmtimedb() entries by key name. } <-- aux_get() fixups. Made
- starttime into a long... time.time() is apparently overflowing int().
-
- 02 Dec 2002; Nicholas Jones <carpaski@gentoo.org> bin/*:
- prep*: Fixed missed man pages and corrected loops to be fairly complete
- in the included files. quickpkg: Loop to do each in $@ (Peter Sharp).
-
- 02 Dec 2002; Nicholas Jones <carpaski@gentoo.org> emerge:
- Yank '++' from searches and make it '\+\+' so that it works. Reformated
- 'all ebuilds are masked' message to be a little more obvious. env_update()
- AFTER autoclean... not in autoclean -- Fixes ldconfig issues. Typo fixes.
-
- 02 Dec 2002; Nicholas Jones <carpaski@gentoo.org> ebuild.sh:
- Added has(), general use()-like function -- 'has this_thing $VAR' returns
- shell true or false. Added has_version(), portage call to determine if
- a specified DEPEND-atom is installed -- 'has_version ">=mozilla-1.2"'
- returns shell true or false. Added best_version(), portage call to
- get the best/most-recently-merged version of a DEPEND-atom statement --
- 'best_version ">=mozilla-1.0"' prints a string. pkg_nofetch() displays
- SRC_URIs by default on RESTRICT -- Can be redefined to display custom
- messages. die() on unpack() failure added. Courtasy of SpanKY: Added
- use_with() and use_enable() for --with-thing and --enable-thing.
- 'use_with gd libgd' would print --with-libgd if gd was in USE and
- --without-libgd if not.
-
- 19 Nov 2002; Martin Schlemmer <azarah@gentoo.org> cnf/etc-update.conf:
- Add 'menu' config item and note about it needing dev-utils/dialog, as
- a lot of users do not know about this nifty feature.
-
- 14 Nov 2002; Phil Bordelon <sunflare@gentoo.org> man/emerge.1:
- Cleaned up the more recent edits to the man page to more closely
- match the previous format, fix various typos, and so on. Expanded
- the REPORTING BUGS section, and made the PACKAGE MASKING part of the
- NOTES a stand-alone section.
-
-*portage-2.0.44 (11 Nov 2002): bug fixes and 1.2 rescue/install setup
-
- 11 Nov 2002; Nicholas Jones <carpaski@gentoo.org> *:
- bin/*: fixes for spaces-in-filename issues, lots of quotes added. conf/*:
- touched up comments and warning. Added RSYNC_RETRIES. ebuild.sh: exit now
- reports the failing ebuild's $CATEGORY/$PF. Added EXTRA_ECONF to econf().
- emake: added EXTRA_EMAKE. emerge: RSYNC_RETRIES curtasy of Christopher
- Sharp. Random comment fixes. portage.py: Random touchups. KV extraction
- fixup for new ROOTs. ARCH missing in profile fix. Fix for packages/All
- missing dir traceback. INFOPATH/INFODIR fix. Added INFOPATH to specials.
- portage.py: Touchups to the mtimedb-touching code. Fixes for the broken
- os.path.normpath() leading '//' bug. Added a modified chuck of Phoen][x's
- dist size patch.
-
- 09 Nov 2002; Nicholas Jones <carpaski@gentoo.org> portage.py:
- Fixups for KV.
-
- 08 Nov 2002; Nicholas Jones <carpaski@gentoo.org> emerge:
- Fix for missing-glibc-bug in --version.
-
- 08 Nov 2002; Nicholas Jones <carpaski@gentoo.org> portage.py:
- Fix for cfgdictfile traceback when ROOT != '/'. Fix for missing kernel
- headers -- checks usr/src/linux, then /usr/include, then if merging
- from sys-kernel ignore error.
-
- 07 Nov 2002; Nicholas Jones <carpaski@gentoo.org> portage.py:
- Fix (part 2) for the USE="-*" arch-missing problem.
-
- 03 Nov 2002; Nicholas Jones <carpaski@gentoo.org> portage.py:
- eclass() function uses a caching and cPickle storing method to determine
- currency of eclass-derived depcache entries. All previous methods are
- already removed. eclass_save() is called via store() now. Stripped some
- cruft from aux_get(). Error message touchups.
-
- 03 Nov 2002; Nicholas Jones <carpaski@gentoo.org> emerge:
- A couple message touchups.
-
- 29 Oct 2002; Nicholas Jones <carpaski@gentoo.org> emerge:
- Moved the autoclean section under the --pretend check so it doesn't flip
- out when running '-f'. Preliminary support for binary-only methods.
- --usepkgonly (implies --usepkg) to force errors is binaries don't exist.
-
- 27 Oct 2002; Daniel Robbins <drobbins@gentoo.org> new*: changed "&&" to "||"
- to provide even better protection against mis-use of these commands.
-
-*portage-2.0.43 (27 Oct 2002): Bug fix release
-
- 27 Oct 2002; Nicholas Jones <carpaski@gentoo.org> emerge:
- AUTOCLEAN="yes" on by default. Added in a clean phase after the merge
- phase so clashing library versions do not remain installed when ldconfig
- is run. This should close up the symlinks-being-removed bugs.
-
- 27 Oct 2002; Nicholas Jones <carpaski@gentoo.org> cnf/*:
- Miscelaneous typo fixes. Added PORTDIR_OVERLAY description in make.conf.
-
- 26 Oct 2002; J Robert Ray <jrray@gentoo.org> portage.py:
- Don't act like the download failed if after successfully downloading
- a file its size doesn't match the file size in the digest. Treat
- this as a mismatched digest condition instead.
-
- 26 Oct 2002; Nicholas Jones <carpaski@gentoo.org> portage.py:
- Reversed the defaults on env_update() so it does a full ldconfig instead
- of a non-symlink ldconfig. Fixed the typo in the 'invalid conf' notices.
-
-*portage-2.0.42 (24 Oct 2002): Bug fix release
-
- 24 Oct 2002; Nicholas Jones <carpaski@gentoo.org> make.defaults.5:
- Removed this file. It is horribly out of date and completely
- wrong to boot. Users shouldn't be modifying the 3 lines in this file
- anyway. They are obvious if you understand conf and defaults.
-
- 24 Oct 2002; Nicholas Jones <carpaski@gentoo.org> emerge:
- Added missing return statements to select_dep() that caused unmet packages
- to be ignored. Touchups to 'emerge info' and 'emerge --version' output.
- Added --delete to --delete-after so that it actually deletes in rsync.
-
- 24 Oct 2002; Nicholas Jones <carpaski@gentoo.org> portage.py:
- Warning touchups. Replaced constants (9) to index("INHERITED").
-
- 23 Oct 2002; Nicholas Jones <carpaski@gentoo.org> bin/new*:
- Fix for "running as a root user" bug... code does 'rm -rf /' if
- variables aren't defined in environment... That's a bad thing.
-
- 23 Oct 2002; Nicholas Jones <carpaski@gentoo.org> portage.py:
- Possible fix for the aux_get() issues and IUSE.
-
- 22 Oct 2002; Nicholas Jones <carpaski@gentoo.org> portage.py:
- noauto got lost in some weird conflicts of .38, added it back.
-
- 22 Oct 2002; Nicholas Jones <carpaski@gentoo.org> emerge:
- Primitive logging via emergelog(). countdown(N,str) function does N second
- countdowns of the action 'str'. Error messages on dependency traceback
- via try/except on select_dep(). Removed the 'unavailable' warning. Message
- updates. Added logging of basic actions to /var/log/emerge.log. Added
- package/to-go counter. 'emerge info' displays a number of useful variables
- that is good for bug reports.
-
- 22 Oct 2002; Nicholas Jones <carpaski@gentoo.org> prepallstrip:
- Bug 9508, fix for MSB architectures strip. Previous check assumed only
- LSB objects could/should be stripped. (*LSB -> *SB) [Joky@#gentoo-sparc]
-
-*portage-2.0.41 (20 Oct 2002): Bug fix release
-
- 20 Oct 2002; Nicholas Jones <carpaski@gentoo.org> portage.py:
- stickies=[] lists the variables that will be kept/used as sticky vars,
- not yet implimented. Fixed several relative symlink bugs by adding
- abssymlink() to portage to determine real/absolute targets. env_update()
- modified to fix bug 9308 symlinks bug. Bug 8348, request for error messages
- instead of tracebacks/dies on parse errors, now caught on a per file basis
- for system config files, includes descriptive messages. Fix for missing
- profile traceback in new prepend'd-arch code. Typo fixes in symlink code.
- DISABLED 'KEYWORD="" == available' code, DEFAULT IS MASKED NOW. Major
- update to masking code, CONFIG_PROTECT can be layered with _MASKs, and
- update_protect() now sets the self.protect* variables. Condensed and
- reordered the unmerge code, now checks existance, cfgprotect, then mtimes.
-
- 20 Oct 2002; Nicholas Jones <carpaski@gentoo.org> emerge:
- Bug 8767: Added 'buildpkg' to FEATURES to specify always-buildpkg.
- Updated the --version output to display profile and gcc version.
- Bug 8083: Eliminate spinner when terminal is not a tty. Bug 7688:
- fixed search vs. searchdesc differences by adding 'cat/pack' split.
- Bug 9308: Symlinks were being killed by ldconfig, fixed by changing
- how/when ldconfig was allowed to set symlinks. rsync command touchups,
- --delete-after so that a failed rsync won't leave a user
- without/with-few packages in portage. More warnings to depclean.
-
- 20 Oct 2002; Nicholas Jones <carpaski@gentoo.org> ebuild.sh:
- Added die to gzip in unpack()
-
- 20 Oct 2002; Nicholas Jones <carpaski@gentoo.org> make.conf*:
- Added in useful variables and descriptions+warnings of their uses.
-
- 20 Oct 2002; Nicholas Jones <carpaski@gentoo.org> make.globals*:
- Added in large "do not edit" message.
-
- 16 Oct 2002; Nicholas Jones <carpaski@gentoo.org> portage.py:
- Fixed an mtimes issue. --update on a package might yank non-cfgprot
- files during the clean-phase because they didn't have mtimes updated.
-
- 15 Oct 2002; Nicholas Jones <carpaski@gentoo.org> portage.py:
- Added 'IUSE' to the auxdbkey list. Fixed random aux_get() errors
- caused by random empty files in the dep cache.
-
- 15 Oct 2002; Nicholas Jones <carpaski@gentoo.org> ebuild.sh
- Actually committed the --host portion of econf.
-
- 15 Oct 2002; Martin Schlemmer <azarah@gentoo.org> portage.py, missingos.c:
- Remove testsandbox.sh, and comment code in portage.py that still
- used it to no real use. Add another '\' to the 'missingos_mknod__doc__'
- string in src/python-missingos/missingos.c to fix failure with gcc-3.3.
-
-*portage-2.0.40 (13 Sep 2002): Bug fix release
-
- 15 Oct 2002; Nicholas Jones <carpaski@gentoo.org> portage.py, emerge:
- emerge depclean -- Removes all packages that are not explicitly or
- dependency merged.
-
- 15 Oct 2002; Nicholas Jones <carpaski@gentoo.org> ebuild.sh:
- depend now prints IUSE. Doesn't do anything with it yet.
- unpack() -- unzip silently overwrites files ().
- econf() -- now specifies '--host=${CHOST}'
-
- 15 Oct 2002; Nicholas Jones <carpaski@gentoo.org> make.globals:
- Touch ups + warning.
-
- 15 Oct 2002; Brandon Low <lostlogic@gentoo.org> etc-update:
- Make etc-update find all the same CONFIG_PROTECT files as portage
- by importing it's settings from portage.
-
-*portage-2.0.39 (13 Sep 2002): Bug fix release
-
- 13 Oct 2002; Nicholas Jones <carpaski@gentoo.org> portage.py:
- Bug 5926 -- If the fetchonly AND pretend flags are specified, then
- portage should give a list of all the SRC_URIs so that they can be
- downloaded or sent into another app. Made sandbox display name in
- 'ps' as '[$PF] sandbox'.
-
- 13 Oct 2002; Nicholas Jones <carpaski@gentoo.org> emerge:
- Bug 5926 -- See portage.py for today. (emerge -pf)
- Typo fixes.
-
- 13 Oct 2002; Nicholas Jones <carpaski@gentoo.org> ebuild.sh:
- Added -q to unzip to make it quiet like the others.
- Bug 6033 -- Fix for infinate loop in eclasses in portage overlay.
-
- 12 Oct 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: fix for
- doebuild() so that our rsync mirror will contain full digests and not just
- partial digests for x86.
-
- 12 Oct 2002; Nicholas Jones <carpaski@gentoo.org> emerge:
- Added --progress to rsync. This puts the file progress meters
- back on. Users have requested this, and it's not detrimental.
- Just creates a lot of output, as the man page says: "It gives
- bored users something to look at."
-
- 12 Oct 2002; Nicholas Jones <carpaski@gentoo.org> portage.py:
- Bug 6361 -- Added checking on directories to ensure that we can
- write to them before we try and traceback. Also added pieces to
- functions so that return conditions are propagated back to emerge
- so that it actually dies on errors. Message on condition details
- restart process.
-
- 11 Oct 2002; Nicholas Jones <carpaski@gentoo.org> make*globals:
- Removed the tomcat configs from CONFIG_PROTECT
-
- 11 Oct 2002; Nicholas Jones <carpaski@gentoo.org> doman:
- Bug 8208 -- Added flags and auto-location support for 'x' manpages.
- Added not-an-man-page message.
-
- 11 Oct 2002; Nicholas Jones <carpaski@gentoo.org> dohtml:
- Bug 8208 -- Added 'js' to the include list, and added '-A' as an
- append flag. '-a' was a filter-down-to.
-
- 11 Oct 2002; Nicholas Jones <carpaski@gentoo.org> portage.py:
- Bug 8964 -- Fix for read-only traceback in digestgen().
- Bug 8949 -- Fix for read-only traceback in doebuild()
- Bug 8284 -- Prepend ARCH _after_ use order processing. Prevents -*
- from killing the ARCH flag. Bug 7596 -- Set an mtime of 0 into
- CONTENTS if the file already exists on the FS. Allows the entry
- to be in the CONTENTS file. Condensed the protect/protect-mask
- code into one function in class dblink. Revised code to allow
- multi-level protects and masks.
-
- 10 Oct 2002; Nicholas Jones <carpaski@gentoo.org> emerge:
- Bug 8552 -- typo fixes
-
- 10 Oct 2002; Mark Guertin <gerk@gentoo.org> cnf/make.conf.ppc :
- Fixed type in reccomended CFLAGS for generic ppc (was 02, corrected
- to O2)
-
- 09 Oct 2002; Nicholas Jones <carpaski@gentoo.org> portage.py:
- Fixed a bug where files less than 2 characters in /etc/env.d
- weren't checked correctly, and tracebacked env_update()
-
- 07 Oct 2002; Martin Schlemmer <azarah@gentoo.org> ebuild.sh:
- Also set $TMP, as MDK among distros sets this, and it causes
- breakage during bootstrap. This should close bug #8101.
-
- 07 Oct 2002; Mark Guertin <gerk@gentoo.org> man/ebuild.5 :
- Updated the man page to include missing functions in portage
-
- 05 Oct 2002; Mark Guertin <gerk@gentoo.org> cnf/make.conf cnf/make.conf.ppc:
- Fixed typo (missing "/") in make.conf and make.conf.ppc, bug #7944
-
- 28 Sep 2002; Daniel Robbins <drobbins@gentoo.org> ebuild.sh: remove
- deprecated "ld.so" from newdepend.
-
-*portage-2.0.38 (25 Sep 2002): Bug fix release
-
- 25 Sep 2002; Daniel Robbins <drobbins@gentoo.org> emerge: now "emerge world"
- and "emerge system" don't replace packages. This closes bug #8282.
-
- 21 Sep 2002; Daniel Robbins <drobbins@gentoo.org> portage-2.0.38.ebuild:
- Added additional perm check for /var/cache/edb/dep dirs and fixed perm
- settings in the ebuild's pkg_postinst(); this should close bug #7719.
-
- 21 Sep 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: Added
- carpaski's patch to add PF to the sandbox's argv[0], closing bug #8141.
-
- 19 Sep 2002; Daniel Robbins <drobbins@gentoo.org> emerge: tweaked emerge so
- that "--pretend" displays don't show "N" all the time.
-
- 19 Sep 2002; Daniel Robbins <drobbins@gentoo.org> ebuild.sh: tweaked help to
- not list deprecated options.
-
- 25 Sep 2002; Maik Schreiber <blizzy@gentoo.org> cnf/make.globals,
- cnf/make.globals.ppc: Added /opt/jakarta/tomcat/conf to CONFIG_PROTECT. Note
- by drobbins: this will be removed soon when this gets moved to an /etc/env.d
- file in the tomcat package.
-
-*portage-2.0.37 (19 Sep 2002): Gentoo Linux 1.4_rc1 version
-
- 10 Sep 2002; Mark Guertin <gerk@gentoo.org> cnf/make.conf.ppc : Backed down
- CFLAGS from -O3 to -O2 on ppc as -O3 has proven to be unreliable on all
- PowerPC-based machines
-
- 09 Sep 2002; Daniel Robbins <drobbins@gentoo.org> emerge: new parameterized
- create() engine, security pass fixes ("emerge" as non-root doesn't print
- "root access required" but shows help instead), emerge --pretend output fixes
- ("to /" lines are dropped and only displayed if installation root != "/".)
- New (and currently unofficial and undocumented) "--deep" and "--selective"
- options that correspond to their respective create() parameters. Using
- "--deep" will enable "deep emerging" -- updating all deps even if the parent
- doesn't need updating. The new parametrized create() is also much cleaner
- than the previous incarnation.
-
- 09 Sep 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: os.chdir() ->
- chdir() fix, enhanced fetch() digest corruption handling.
-
- 06 Sep 2002; Mark Guertin <gerk@gentoo.org> cnf/make.globals.ppc :
- Updated make.globals.ppc to reflect new CFLAGS for 1.4 release
-
-*portage-2.0.36 (04 Sep 2002): Gentoo Linux 1.4_rc1 version
-
- 04 Sep 2002; Daniel Robbins <drobbins@gentoo.org> emerge: "--onlydeps" was
- mostly broken since the most recent emerge code restructure. Now fixed,
- closing bug #7442.
-
- 04 Sep 2002: Daniel Robbins <drobbins@gentoo.org> etc-update.conf: added
- missing etc-update config file.
-
- 04 Sep 2002: Daniel Robbins <drobbins@gentoo.org> portage.py: removed error
- detection based on return value for pkg_preinst() and friends when called
- from merge() and unmerge(), since the value can be unreliable due to the &&
- shell construct, depmod -a returning a non-zero value, etc. Also fixed
- env_update() to treat CONFIG_PROTECT and CONFIG_PROTECT_MASK correctly.
-
-*portage-2.0.35 (03 Sep 2002)
-
- 03 Sep 2002: Daniel Robbins <drobbins@gentoo.org> etc-update,
- man/etc-update.1: new program for updating config files.
-
- 03 Sep 2002; Daniel Robbins <drobbins@gentoo.org> ebuild.sh: upgraded
- inherit() to support $PORTDIR_OVERLAY, closing bug #6033.
-
- 03 Sep 2002; Daniel Robbins <drobbins@gentoo.org> emerge: calls to "cvs" now
- use "-z3" compression option, closing bug #5982.
-
- 03 Sep 2002; Daniel Robbins <drobbins@gentoo.org> ebuild.sh: symlinks to
- .tbz2's in the $PKGDIR/All directory (from $PKGDIR/$CATEGORY) are now
- relative, closing bug #6881.
-
- 03 Sep 2002; Daniel Robbins <drobbins@gentoo.org> ebuild.sh: einstall now
- sees "GNUmakefile", closing bug #4895.
-
- 03 Sep 2002; Daniel Robbins <drobbins@gentoo.org> ebuild.sh: typo fixes,
- closing bug #7263.
-
- 03 Sep 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: /var/tmp
- creation now works even if /tmp already exists, closing bug #7376.
-
- 02 Sep 2002; Daniel Robbins <drobbins@gentoo.org> emerge: fixed a logic error
- that resulted in non-root emerge failure. This closes bug #7389.
-
-*portage-2.0.34 (01 Sep 2002)
-
- 01 Sep 2002; Daniel Robbins <drobbins@gentoo.org>: portage.py, ebuild.sh:
- Azarah's ebuildsh() removed, various other little fixes like pkg_preinst and
- friends not dying on non-zero return codes.
-
- 01 Sep 2002; Martin Schlemmer <azarah@gentoo.org> ebuild.sh: add a custom
- version of the esyslog() function to fix the "Red Star" bug if no logger is
- running.
-
-*portage-2.0.33_p1 (30 Aug 2002): urgent fix
-
- 31 Aug 2002; Dan Armak <danarmak@gentoo.org> ebuild.sh: comment out the if
- clause that only executed inherit() conditional on $PORTAGE_RESTORE_ENV. It
- didn't work and broke portage 2.0.33 as far as inheriting ebuilds goes.
-
-*portage-2.0.33 (30 Aug 2002)
-
- 30 Aug 2002; Daniel Robbins <drobbins@gentoo.org> : some additional clean-ups
- for the make.conf(.ppc) files.
-
- 30 Aug 2002; Daniel Robbins <drobbins@gentoo.org> emerge: emerge search would
- die when aux_get() raised a KeyError; we now catch and handle this exception.
- This closes bug #7280.
-
- 29 Aug 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: fixed typo in
- move_ent() code that messed up moving packages into previously non-existant
- category directories in /var/db/pkg.
-
- 30 Aug 2002; Mark Guertin <gerk@gentoo.org> cnf/make.conf.ppc : Updated
- cnf/make.conf.ppc with new CFLAGS for 1.4 release
-
-*portage-2.0.32 (29 Aug 2002)
-
- 29 Aug 2002; Daniel Robbins <drobbins@gentoo.org> emerge: favorites weren't
- working since 2.0.30; they're working again now. This closes bug #7225.
-
- 29 Aug 2002; Phil Bordelon <sunflare@gentoo.org> emerge.1: Added a small
- blurb regarding what to do when emerge --update [world|system] fails because
- of new features such as || and ?. Hopefully this will assuage the fairly
- common questions regarding this.
-
-*portage-2.0.31 (29 Aug 2002)
-
- 28 Aug 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: fixed a config
- file protection logic bug detected and identified by Azarah. It basically
- broke most config file updates and has been broken for around six months.
- Very surprised no one had encountered this issue before.
-
- 28 Aug 2002; Daniel Robbins <drobbins@gentoo.org> emerge: improved error
- messages when an unsatisfiable dependency is encountered. Now lets you know
- if there are masked packages that would satisfy the dep.
-
- 28 Aug 2002; Daniel Robbins <drobbins@gentoo.org> emerge: emerging .tbz2
- packages and .ebuilds by name now works; this was broken in 2.0.30.
-
-*portage-2.0.30 (28 Aug 2002)
-
- 28 Aug 2002; Daniel Robbins <drobbins@gentoo.org> portage.py, emerge: rewrote
- gettimeval() as cpv_counter() and updated it to handle corrupt COUNTER files,
- closing bug #6763.
-
- 28 Aug 2002; Daniel Robbins <drobbins@gentoo.org> emerge: "emerge search" now
- displays information for masked packages if possible, closing bug #6823.
-
- 28 Aug 2002; Daniel Robbins <drobbins@gentoo.org> emerge: fixed an old gbevin
- bug where "emerge search" wouldn't show packages whose version string was
- only one character long, ie. "foo-3". This closes bug #6800.
-
- 28 Aug 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: added a missing
- call to flatten() in dep_check(), allowing recursive sublists in dependencies
- and SRC_URI variables to work; this closes bugs #7104, #7116, #7122.
-
- 28 Aug 2002; Daniel Robbins <drobbins@gentoo.org> doman: fixed some typos,
- closing bug #7152.
-
- 28 Aug 2002; Daniel Robbins <drobbins@gentoo.org> portage.py, emerge: major
- work to emerge: reworked the internal code organization to make depgraph code
- easier to understand; unified two code paths so that ebuild/.tbz2 choices
- when "--usepkg" should now be eternally consistent; tweaked portage.py so
- that dep_expand()'s dbapi argument is optional. In the process of all this,
- isolated and fixed bug #4508.
-
- 27 Aug 2002; Martin Schlemmer <azarah@gentoo.org> portage.py: if the dep
- cache is stale, it is possible that aux_get() will call doebuild(depend) to
- regenerate it again. This call will cause $T to be set to "", which will
- break anything that needs $T to be set to a writable location inside the
- sandbox, so we need to set $T to a valid value again.
-
- 27 Aug 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: improved error
- handling/detection of bad pkgsplit() and catpkgsplit() calls; should close
- bug #6803, #6853.
-
- 26 Aug 2002; Daniel Robbins <drobbins@gentoo.org> emerge: fix so that "emerge
- prune rsync" doesn't complain that you are trying to rsync and prune at the
- same time. This closes bug #6785.
-
- 26 Aug 2002; Daniel Robbins <drobbins@gentoo.org> doman fixes; should work
- for pre-gzipped man-pages and man-pages with multiple "."s in their name.
- Closes bugs #6770, #6917
-
- 25 Aug 2002; Martin Schlemmer <azarah@gentoo.org> emerge: fixed an
- indentation problem of the 'else:' at line 1304.
-
- 20 Aug 2002; Daniel Robbins <drobbins@gentoo.org> emerge: removed stray
- "DEBUG:" output, closing bug #6732.
-
-*portage-2.0.29 (18 Aug 2002)
-
- 18 Aug 2002; Daniel Robbins <drobbins@gentoo.org> emerge, portage.py,
- ebuild.sh: making some effort to make output cleaner and less cluttered,
- particularly with error handling.
-
- 18 Aug 2002; Daniel Robbins <drobbins@gentoo.org> emerge: updated so that
- "emerge --inject foo-1.0" fails instead of injecting "null/foo-1.0"; fixed
- error in help; "emerge --inject" and "emerge" with no specified files or
- package classes prints out a small warning and exits. Closes bug #6353.
-
- 18 Aug 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: movefile()
- didn't like bind mounts due to a peculiar quality -- according to their
- ST_DEV stat() information, they are on the same filesystem, so rename()
- should work, but it doesn't. We now fall back to copy if rename() doesn't
- work. Closes bug #6468.
-
- 18 Aug 2002; Daniel Robbins <drobbins@gentoo.org> ebuild.sh: made significant
- enhancements and fixes to ebuild.sh's error-handling code. Errors in
- src_unpack(), src_compile(), src_install() and others should now be correctly
- detected. Also downgraded our use() function since we are not implementing
- extended USE functionality. Closes bug #6393.
-
- 18 Aug 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: portdb's
- cp_all() now skips "CVS" directories, closing bug #6662.
-
- 18 Aug 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: aux_get():
- exception handler to print informative message when encountering wacky cache
- entries that we just can't fix (for some reason.)
-
- 17 Aug 2002; Daniel Robbins <drobbins@gentoo.org> ebuild.sh: removed
- deprecated "ld.so" dependency from Dan Armak's newdepend() function.
-
- 17 Aug 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: doebuild()
- running in "depend" mode will set $T to "" to prevent a parent's $T from
- being inherited; this is important when we run custom portage code from
- inside an ebuild (such as pkg_postinst)... it allows dep caching to not
- break. Closes bug #6484.
-
-*portage-2.0.28 (17 Aug 2002)
-
- 16 Aug 2002; Daniel Robbins <drobbins@gentoo.org> emerge: return errors when
- two actions like "world" and "system" are specified on the command-line, or
- when "system" or "world" are combined with package names. Closes bug #6492.
-
- 16 Aug 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: bad string was
- causing pkg_prerm() and pkg_postrm() to not be called; now fixed. This
- closes bug #6493.
-
- 16 Aug 2002; Daniel Robbins <drobbins@gentoo.org> emerge: now calls
- env-update() after an unmerge() run (to fix library paths, links, etc.)
- Closes bug #6511.
-
- 16 Aug 2002; Daniel Robbins <drobbins@gentoo.org> doman: now handles already-
- gzipped man pages correctly, closing bug #6544.
-
- 16 Aug 2002; Daniel Robbins <drobbins@gentoo.org> emerge: fixed a bug where a
- dep specified on the command-line with --usepkg would use an old package if
- it happened to match the dep. Now, it will use the ebuild instead, just like
- how deps of deps are treated.
-
-*portage-2.0.27 (06 Aug 2002)
-
- 07 Aug 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: removed some
- beta extended USE functionality to speed up regenerate(); as we are likely
- not going to need extended USE for a good while.
-
- 07 Aug 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: ebuild.sh now
- supports "config" (pkg_config) again.
-
- 06 Aug 2002; Dan Armak <danarmak@gentoo.org> ebuild.sh: change debug-print()
- to exit if $T is not defined, i.e. if emerge is running in dependency
- detection mode and isn't actually emerging everything. This fixes bug #4932
- (the "eclass-debug.log is created in /" problem).
-
- 06 Aug 2002; Nicholas Jones <carpaski@gentoo.org> ebuild.sh: Fixed a
- case-check problem that prevented tar.Z from being un-tar'd. This closes bug
- #6126.
-
- 06 Aug 2002; Martin Schlemmer <azarah@gentoo.org> portage.py, ebuild.sh:
- Remove "setup" from 'sandboxactive' in portage.py, as we are not running
- pkg_setup() in a sandbox anymore. Move the restoring of the old env code
- to be the first thing in ebuild.sh, just to ensure we dont mess anything
- up when greping/awking during saving it.
-
-*portage-2.0.26 (06 Aug 2002)
-
- 06 Aug 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: on unmerge,
- entries are now removed from the world file if 1) the dependency refers to
- the cat/pkg being unmerged, and 2) the dependency matches the current version
- being unmerged, and 3) unmerging this package will leave no other packages on
- this system that will match this world entry. This closes bug #3409.
-
- 06 Aug 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: new global
- update support. You can now add an entry to a file in
- /usr/portage/profiles/updates/, something like "move x11-base/xfree
- x11-base/xfree86". Portage will then update the /var/db/pkg db on user's
- boxes so that any xfree packages are renamed to xfree86; it will also update
- the world and virtuals files appropriately. We are naming the update files
- "3Q-2002", etc. -- for third quarter in 2002. This way, we don't bog
- Portage down by having it run through all our directives. It will only look
- at update files whose mtimes have changed. Closes bug #4753, #5463,
-
- 06 Aug 2002; Daniel Robbins <drobbins@gentoo.org> ebuild.sh: tweaked Azarah's
- environment-saving patch to not be enabled when $T is not defined -- this
- closes bug #6070 and should speed up dep calculations too. Also tweaked
- ebuild.sh so that pkg_setup() is run *outside* of the sandbox. This is
- important because piping stuff around tends to require temp files to be
- created; pkg_setup() being in the sandbox prevents this from happening.
-
- 06 Aug 2002; Martin Schlemmer <azarah@gentoo.org> ebuild.sh : Unset
- esave_ebuild_env in esave_ebuild_env(), otherwise the sourced copy messes
- saving of the environment.
-
- 05 Aug 2002; Phil Bordelon <sunflare@gentoo.org> emerge.1: Updated the man
- page to match the current version of 2.0.25.
-
- 05 Aug 2002; Phil Bordelon <sunflare@gentoo.org> emerge.1: Updated the man
- page to match the current version of 2.0.25.
-
-*portage-2.0.25 (05 Aug 2002)
-
- 05 Aug 2002; Daniel Robbins <drobbins@gentoo.org> src/sandbox/Makefile:
- removed -march=i386 added in Azarah's patch; breaks things for PPC, Sparc.
-
- 05 Aug 2002; Daniel Robbins <drobbins@gentoo.org> emerge: logic error fix for
- description searching, closing bug #6031.
-
-*portage-2.0.24 (04 Aug 2002)
-
- 04 Aug 2002; Daniel Robbins <drobbins@gentoo.org> output.py, emerge: rewrote
- our output.py module (it was weird) and tweaked emerge so that colorization
- will be disabled if NOCOLOR is set to "yes" or "true" *or* if sys.stdout isn't
- a tty. This means that if you pipe things to "less", colorization will get
- automatically disabled. Yay! This closes bug #5714,
-
- 04 Aug 2002; Daniel Robbins <drobbins@gentoo.org> ebuild.sh: default
- src_compile now properly uses "die" instead of "return 1." This closes bug
- #2981.
-
- 04 Aug 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: "=foo/bar-1.0*"
- now matches "foo/bar-1.0_{alpha|beta|pre|rc}{int}" but not
- "foo/bar-1.1_{alpha_beta_pre_rc}{int}". This closes bug #5874; gcc-3.2_pre
- is now considered a "3.2" rather than a very late "3.1" when doing "*"
- matching.
-
- 04 Aug 2002; Daniel Robbins <drobbins@gentoo.org> ebuild.sh: Applied
- carpaski's patch to allow "unpack" to also handle regular .gz or .bz2 files;
- in which case the unpacked files are placed directly in ${WORKDIR}. Closing
- bug #5867.
-
- 04 Aug 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: tweaked
- exithandler() to only do its sandbox clean-up if we happen to be the root
- user. Closes bug #5859.
-
- 04 Aug 2002; Daniel Robbins <drobbins@gentoo.org> emerge: search now lists
- masked packages correctly. Closes bug #5854 (pending QA verification.)
-
- 04 Aug 2002; Phil Bordelon <sunflare@lsu.edu> emerge.1: The emerge man page
- now reflects the fact that emerge sync|rsync always does a --clean now. Also
- bumped the release number to 2.0.23.
-
- 03 Aug 2002; Martin Schlemmer <azarah@gentoo.org> portage.py, emerge.sh,
- sandbox.c libsandbox.c: Updated sandbox.c and libsandbox.c to use an internal
- env variable, $SANDBOX_ACTIVE to determine if sandbox should really be active
- or not. With it only checking $SANDBOX_ON, some instances NOT running in a
- sandbox, but that set SANDBOX_ON while an actual sandbox was running, caused
- the sandbox to activate for this process. Added in support for a more phased
- calls to ebuild.sh again. This is this time done with support to save the
- current environment of ebuild.sh to the next call that should handle problems
- with pkg_setup() setting env variables. Closes bugs #5853, #5817, #5950.
-
-*portage-2.0.23 (01 Aug 2002)
-
- 01 Aug 2002; Daniel Robbins <drobbins@gentoo.org> portage.py:
- ${ROOT}var/cache/edb and friends will get created if they don't exist,
- closing bug #5813.
-
- 01 Aug 2002; Daniel Robbins <drobbins@gentoo.org> emerge: rewrote the emerge
- search code, making it fully API-compliant and much more streamlined and
- compact. This rewrite should fix the problem where emerge search doesn't
- support Portage overlays (bug #5783.)
-
- 31 Jul 2002; Phil Bordelon <sunflare@gentoo.org> emerge.1: Documented the
- fact that emerge clean does not remove unslotted ebuilds. Bumped the man
- page revision number to 2.0.22.
-
-*portage-2.0.22 (29 Jul 2002)
-
- 29 Jul 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: fetch() didn't
- handle resuming downloads properly when the first attempt aborted
- prematurely. The fall back to the alternate location would not resume the
- download. This should now be fixed. Closes bug #5655.
-
- 29 Jul 2002; Daniel Robbins <drobbins@gentoo.org> portage.py, ebuild.sh: I
- had some code in spawn() from one of Azarah's patches that I forgot to
- remove; it caused the sandbox to be disabled all the time. I removed this
- code, and then I tweaked ebuild.sh so that the sandbox runs in a "deny by
- default" configuration. Before, you could add lines to the main ebuild
- (outside of a function) and it would bypass the sandbox; no more. Closes
- bugs #5740, #5744.
-
- 29 Jul 2002; Daniel Robbins <drobbins@gentoo.org> emerge: --help and
- --version now work for non-root users as they should. --clean has been
- deprecated. This closes bug #5658.
-
-*portage-2.0.21 (28 Jul 2002)
-
- 28 Jul 2002; Daniel Robbins <drobbins@gentoo.org> portage.py, emerge: Added
- beta "Portage tree overlay" support. By setting PORTDIR_OVERLAY to point to
- a local directory tree, you can cause Portage to look for ebuilds in
- PORTDIR_OVERLAY first before consulting the regular PORTDIR. Using this
- feature, it's possible to have your PORTDIR set up to rsync but still be able
- to have locally-created ebuilds in your PORTDIR_OVERLAY tree. For example,
- you would place your ebuild in PORTDIR_OVERLAY/sys-apps/foo/.
-
- 27 Jul 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: another
- dep_opconvert() USE-handling bug-fix. Working OK for Azarah now.
-
- 27 Jul 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: small logic
- error fix in dep_opconvert() to fix up USE handling.
-
- 27 Jul 2002; Daniel Robbins <drobbins@gentoo.org> emerge: security modes now
- work correctly; root access now required for merging. The new security pass
- code was missing a check that is now present.
-
-*portage-2.0.20 (27 Jul 2002)
-
- 27 Jul 2002; Daniel Robbins <drobbins@gentoo.org> portage.py, emerge,
- repoman: Rewrote a good deal of our dependency-checking code, including all
- of dep_opconvert(). Results? repoman will now auto-enable all USE variables
- (even ! use variables will get enabled) resulting in thorough checks of all
- specified dependencies. DEPEND="foo? bar : oni" and DEPEND="foo? ( bar oni )
- : ( meep barf )" now works correctly. DEPEND="|| ( foo bar oni )" now works
- correctly and will try to satisfy "foo" (the first package) if none are
- installed. DEPEND="!foo? ( bar )" now works correctly; it was not working
- before. This should generally mean that our dependency system is now working
- as expected.
-
- 27 Jul 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: changed
- counter_tick() method so that bulk of code is in the counter_tick_code()
- helper function; then created a new method for fakedbapi that calls
- counter_tick(). This fixes "--emptytree" issues with emerge, since emerge
- still expects counter_tick() to exist as a method.
-
- 27 Jul 2002; Daniel Robbins <drobbins@gentoo.org> portage.py, emerge: removed
- a DEBUG: print and a stray "raise IndexError" that I used for debugging.
-
-*portage-2.0.19 (26 Jul 2002)
-
- 26 Jul 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: Portage is now
- eclass-friendly when it comes to regenerating cache entries. This closes bug
- #4843.
-
- 26 Jul 2002; Daniel Robbins <drobbins@gentoo.org> portage.py, emerge: Closed
- bug #5524. Packages "emerge inject"ed will no longer get auto-cleaned
- indiscriminantly. Injected packages now have an official COUNTER so that
- emerge clean doesn't get confused and schedule the package for removal. Also
- modularized and improved the counter-handling functions and changed the
- counter update method so that systems with XFS filesystems that die
- unexpectedly will no longer get corrupted COUNTER files. This should solve
- the an entire class of "my counter is corrupt" issues for XFS users.
-
- 26 Jul 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: added exception
- handler to getcontents() so that it will gracefully ignore (with a warning)
- corrupt CONTENTS file lines. This closes bug #5464.
-
- 26 Jul 2002; Daniel Robbins <drobbins@gentoo.org> emerge: fixed a single-line
- indentation error that resulted in "emerge clean" ignoring some files. This
- closes bug #5597,#4364.
-
- 25 Jul 2002; Daniel Robbins <drobbins@gentoo.org> emerge: --noreplace is now
- off by default and works again; --onlydeps appears to be working again; and
- merge() will merge things as specified in display(), which was not the case
- before (I rolled 2.0.18 before fixing a few things I forgot about)
-
-*portage-2.0.18 (25 Jul 2002)
-
- 25 Jul 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: Close bug
- #5580; pkg_setup() now gets called during all build-related stages.
-
- 24 Jul 2002; Daniel Robbins <drobbins@gentoo.org> portage.py, emerge: rewrote
- create() function, closing bug #5469. Also added freeze() and melt() methods
- to portdbapi to enhance performance. create() is now extensively commented
- and even understandable.
-
- 24 Jul 2002; Daniel Robbins <drobbins@gentoo.org> emerge: the "rsync --clean"
- option has been deprecated. Cleaning is now on by default, closing bug
- #5527.
-
- 24 Jul 2002; Daniel Robbins <drobbins@gentoo.org> portage.py, make.globals:
- It should now be safe to do what Spider did in the gconf ebuild and add
- CONFIG_PROTECT and CONFIG_PROTECT_MASK to /etc/env.d entries. Both variables
- are now incremental, just like USE, and you can use "-path" to turn a path
- off, or add CONFIG_PROTECT{_MASK}="path" to *add* a path to the list (this
- will not overwrite "parent" settings. Also added "/etc/env.d/" to
- CONFIG_PROTECT_MASK in /etc/make.globals. Also, very importantly, /etc/env.d
- is now hard-coded into CONFIG_PROTECT_MASK as it was in earlier versions of
- Portage and cannot be removed.
-
- 24 Jul 2002: Daniel Robbins <drobbins@gentoo.org> emerge, portage.py: two
- fixes; first, I removed xcache.p support from portage.py. The code was
- designed based on the false theory that directory mtimes are updated whenever
- an object inside that directory is modified. That is not the case -- it
- happens when the directory listing itself changes. Also fixed overly verbose
- emerge merging error.
-
- 23 Jul 2002; Phil Bordelon <sunflare@gentoo.org> emerge.1: Updated the man
- page to match some changes in the latest version of portage, along with some
- fixes recommended by drobbins.
-
-*portage-2.0.17 (23 Jul 2002)
-
- 23 Jul 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: fixed a bug
- pointed out by trance -- dep_nomatch() was still using the old (deleted)
- match() function. Now it's using a new match() method.
-
-*portage-2.0.16 (22 Jul 2002)
-
- 22 Jul 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: changed xcache
- behavior so that the cache gets totally thrown away when the masks are
- updated. Also revamped carpaski's code so we can avoid loading xcache from
- disk if we simply plan to throw it away. This necessitated the creation of a
- new file in /var/cache/edb: mtimes. This file stores mtimes for various
- important filesystem objects. Also added support to not regenerate the GNU
- info directory index if the mtime on /usr/share/info has not changed since
- previous invocation.
-
- 22 Jul 2002; Daniel Robbins <drobbins@gentoo.org> portage.py, emerge: moved
- security handling code from emerge to portage.py; also moved group "wheel"
- check to portage.py. Added additional security checks and permissions fixes
- to the cache handling functions.
-
- 22 Jul 2002: Daniel Robbins <drobbins@gentoo.org> emerge: fixed a bug
- reported by woodchip related to merging packages. When using --usepkg, a
- package wouldn't be used if there was a newer unmasked version of the ebuild
- available, even if the package in question was the right selection.
-
- 22 Jul 2002; Daniel Robbins <drobbins@gentoo.org> emerge: fixed create() code
- and consolidated some functions. My goal is to simplify the emerge code
- until it's clean, at which point we can begin adding new features to emerge
- like more SLOT-friendly decision making.
-
- 22 Jul 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: was able to
- remove 130 or so lines of code by depreciating the match() function (replaced
- by the match2() method.) I also rennovated the binarytree code to take
- advantage of fakedbapi.
-
- 22 Jul 2002; Nicholas Jones <carpaski@gentoo.org> portage.py: added the
- xcache.p fix and updated the version to 2.0.16pre (from 2.0.12) -- Also added
- in a try/except inside the store() (atexit) for xcache.p to give hints
- instead of tracebacks.
-
- 22 Jul 2002; Nicholas Jones <carpaski@gentoo.org> emerge: added the
- description searching code and added a spinner to the search.
-
- 21 Jul 2002; Phil Bordelon <sunflare@gentoo.org> emerge.1: Finished the
- rewrite of the emerge man page.
-
- 21 Jul 2002; Phil Bordelon <sunflare@gentoo.org> emerge: Readded the sync
- help to the list of help options now that emerge --help sync works again.
-
-*portage-2.0.15 (16 Jul 2002)
-
- 15 Jul 2002; Daniel Robbins <drobbins@gentoo.org> emake: no longer falls back
- to plain old "make" if parallel make fails. I believe it's best to fix the
- problem (turn parallel make off in the ebuild) rather than tweak emake to
- avoid it.
-
- 15 Jul 2002; Daniel Robbins <drobbins@gentoo.org> emerge: rewrote some code,
- cleaning things up and removing redundant functions. Starting to
- de-cruftify.
-
- 15 Jul 2002; Daniel Robbins <drobbins@gentoo.org> emerge: fix for permissions
- when updating the cache using server-generated entries.
-
-*portage-2.0.14 (15 Jul 2002)
-
- 15 Jul 2002; Daniel Robbins <drobbins@gentoo.org> ebuild.sh: ccache bug fix;
- removed extra "!". Ccache should now get enabled properly. Also, INHERITED
- variable is now getting added to the dep cache entries for future
- eclass-friendly caching.
-
- 15 Jul 2002: Daniel Robbins <drobbins@gentoo.org> portage.py: Portage now
- *persistently* caches xmatch() calculations. This appears to speed things up
- but we will need to limit the size of the new xcache.p in future versions of
- Portage.
-
- 15 Jul 2002; Daniel Robbins <drobbins@gentoo.org> portage.py, emerge: Portage
- now has keyword-based masking enabled. It also has server-side caching
- enabled to eliminate "emerge pre-Calculating dependencies... delay" for
- end-users. There is now an undocumented "regen" option for emerge that tells
- Portage to ensure that all entries in /var/cache/edb/dep are up-to-date.
- This is intended mainly for developers, as the server-side caching feature
- should ensure that all dep cache entries are up-to-date for end-users.
-
- 15 Jul 2002; Daniel Robbins <drobbins@gentoo.org> repoman: repoman now checks
- for ebuilds that generate output as well as ebuilds that return a non-zero
- error code when sourced. For this test to detect all failures, you need to
- wipe out your /var/cache/edb/dep/* before running repoman. This design quirk
- is necessary to preserve existing emerge behavior.
-
- 13 Jul 2002: Daniel Robbins <drobbins@gentoo.org> portage.py: Removed
- Azarah's ebuildsh() function as it breaks pkg_setup() again. pkg_setup()
- needs to be called as part of the same process as any other ebuil.sh command
- so that environment vars set in pkg_setup() are preserved through the rest of
- the build process.
-
- 13 Jul 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: changed calling
- convention for doebuild("depend").
-
- 13 Jul 2002; Daniel Robbins <drobbins@gentoo.org> ebuild.sh: removed
- erroneous "local dbkeys" that was outside of a function. This fixes some
- cache update problems. Added INHERIT to our cache entries so we can
- correctly update the cache for eclass ebuilds.
-
- 12 Jul 2002; Dan Armak <danarmak@gentoo.org> make.globals, make.globals.ppc:
- remove default KDE2DIR, KDE3DIR settings and comments. The kde eclasses will
- now handle the case where they are not defined. This will be used with kde
- 3.1 and later, and the difference between their default value and them not
- being defined is important.
-
-*portage-2.0.13 (11 Jul 2002)
-
- 11 Jul 2002; Daniel Robbins <drobbins@gentoo.org> tarball.sh: our tarball
- script didn't clean the src/sandbox directory before creating our distribution
- tarball. This resulted in src/sandbox/sandbox (the executable) being distributed,
- and the "make" in the ebuild thinking everything was up-to-date. The result?
- Everyone in the world got a sandbox compiled with gcc 3.1. This has been
- fixed, closing bug #4867, #4851.
-
- 10 Jul 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: removed some
- bogus code that snuck in. pkg_setup() is no longer called when installing a
- tbz2. All tbz2 stuff should be done in pkg_pre/postinst().
-
-*portage-2.0.12 (10 Jul 2002)
-
- 10 Jul 2002; Daniel Robbins <drobbins@gentoo.org> ccache support is now
- controlled via a "ccache" FEATURES variable, which is enabled in make.globals
- my default. ccache support can now be turned off by adding a
- FEATURES="-ccache" to /etc/make.conf. Also, CCACHE_DIR correctly detected
- and utilized.
-
- 10 Jul 2002; Daniel Robbins <drobbins@gentoo.org> emerge: "emerge --help
- rsync" now works correctly and displays rync help rather than rsyncing. This
- closes bug #4438, #4629.
-
- 10 Jul 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: Portage now
- prints a friendly error if PORTAGE_TMPDIR doesn't exist or is not a
- directory. This effectively closes bug #4360.
-
- 10 Jul 2002; Daniel Robbins <drobbins@gentoo.org> emerge: the "inject",
- "sync" and "rsync" actions no longer allow "--pretend" or "-p" to be
- specified. This closes bug #4352.
-
- 10 Jul 2002; Daniel Robbins <drobbins@gentoo.org> sandbox.c: use realpath()
- to expand the PORTAGE_TMPDIR, /var/tmp, /tmp paths. This allows write access
- to these directories even if /var or /var/tmp is a symlink, for example.
- Without this fix, access to these directories will be denied by the sandbox,
- creating a bunch of problems. Closes bugs #4256, #2379, #4625, #2931, #4829.
-
- 10 Jul 2002; Daniel Robbins <drobbins@gentoo.org> megadigest, megadownload,
- megatouch, pkgsearch: removed from the bin/ directory; deprecated.
-
- 10 Jul 2002; Daniel Robbins <drobbins@gentoo.org> emerge, quickpkg: remove
- hard-coded references to "/usr/portage" in "emerge search", quickpkg. Now
- correctly uses PORTDIR instead. Closes bug #4836.
-
- 10 Jul 2002; Daniel Robbins <drobbins@gentoo.org> emerge: emerge now checks
- to see if the "wheel" group exists before running; if it doesn't, it exits
- with a polite error message. Closes bug #4736.
-
- 08 Jul 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: changed the
- unmerge() code so that it doesn't use mtimes to test whether a symlink
- should be unmerged, since mtimes and symlinks are weird partners. Instead,
- we save unmerging of our symlinks until the end of the code, and unmerge
- them only if their target no longer exists. This closes bug #4491.
-
- 08 Jul 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: check whether
- ebuild actually exists immediately before running pkg_postint() and
- pkg_preinst() from the dblink merge() method. Also change all PKG_TMPDIR
- references to PORTAGE_TMPDIR plus suffix, closing bug #4447, #4853.
-
-*portage-2.0.11 (07 Jul 2002)
-
- 07 Jul 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: rewrote
- portdbapi's xmatch() and visible() to fix significant bugs. Logic errors in
- both methods caused ~ deps to not work correctly. After the rewrite,
- dependency checking is now 44% faster.
-
- 07 Jul 2002; Daniel Robbins <drobbins@gentoo.org> repoman, portage.py: Added
- new DEPEND and RDEPEND.badmasked categories to repoman. repoman now checks
- dependencies of masked packages using *all* ebuilds, rather than trying to
- match them against all visible ebuilds. DEPEND.bad and RDEPEND.bad (used to
- be ".unsolvable") now only tally visible ebuilds, not masked ones.
-
-*portage-2.0.10 (06 Jul 2002)
-
- 05 Jul 2002; Daniel Robbins <drobbins@gentoo.org> portage.py, repoman: Added
- new capabilities to repoman -- the ability to detect unsolvable DEPEND and
- RDEPEND variables, missing DESCRIPTION, LICENSE, KEYWORDS and SLOT. Fixed
- repoman so that adding a comment with quotes in it doesn't break things.
- Added a few tiny extensions to portage.py to support the new repoman
- features.
-
- 03 Jul 2002; Martin Schlemmer <azarah@gentoo.org> ebuild.sh: some ebuilds
- like gcc do not use $S to build the package in, and this causes generated .la
- files (libtool) to contain $WORKDIR in them. We thus export $WORKDIR in
- ebuild.sh as well, to go along with fixes to the libtool-portage patch.
-
- 02 Jul 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: pkg_setup() now
- gets called before a tbz2 is installed, and pkg_preinst() and postinst() get
- called at the right times as well.
-
- 01 Jul 2002; Phil Bordelon <sunflare@gentoo.org> emerge: "emerge --help
- rsync" does an rsync instead of printing help. Until this is fixed, I
- removed it from the list of detailed help options.
-
-*portage-2.0.9 (01 Jul 2002)
-
- 01 Jul 2002; Daniel Robbins <drobbins@gentoo.org> Added missing .match()
- method to portdbapi.
-
- 01 Jul 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: Added new
- operator for dependencies: DEPEND="foo? bar : oni" will use oni if foo isn't
- set. Added "||" support back after removing it from my working copy; looks
- ok.
-
- 29 Jun 2002; Daniel Robbins <drobbins@gentoo.org> portage.py, make.globals,
- make.globals.ppc: internalized definitions of BUILDDIR and PKG_TMPDIR in
- order to make things work as expected when one sets PORTAGE_TMPDIR. The new
- config file var expansion algorithm made this change necessary.
-
- 29 Jun 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: fix doebuild()
- so that pkg_setup() gets called when a .tbz2 package is being built. This
- closes bug #3673.
-
- 29 Jun 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: tweaked the
- aux_get() code to automatically regenerate cache files if they have the
- incorrect number of entries. Added an additional fix to this code on 01 Jul
- 2002.
-
- 29 Jun 2002; Martin Schlemmer <azarah@gentoo.org> portage.py: merge in some
- of the missing ld.so.preload fixes again.
-
- 29 Jun 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: new version
- with highly-optimized dependency calculation caching, particularly for
- portdbapi. Includes new three-level caching portdbapi xmatch() method.
-
- 27 Jun 2002; Grant Goodyear <g2boojum@gentoo.org> ebuild.sh: Added keepdir()
- function so that it no longer has to be hardcoded in ebuilds.
-
-*portage-2.0.8 (27 Jun 2002)
-
- 27 Jun 2002; Daniel Robbins <drobbins@gentoo.org> emerge: in an attempt to
- provide a predictable release of Portage, I'm reverting the "emerge
- --pretend" "fix" in Portage-2.0.6. It ignores custom USE settings when
- calculating child deps, which makes things tricky for users. We'll use the
- "expected" (old) behavior for now until we have a solution for this USE
- issue.
-
- 27 Jun 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: fixed handling
- of ! deps in match(), closing bug #4219. Thanks Spidler!
-
- 27 Jun 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: disabled USE
- regeneration on reset() for performance purposes.
-
-*portage-2.0.7 (26 Jun 2002)
-
- 26 Jun 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: add a friendly
- error handler for aux_get() so that flaky ebuilds don't cause it to trip up
- with a cryptic traceback; users will get a friendly error message instead.
- Also, temporarily disable keyword-based masking, since it currently slows
- down Portage by quite a bit since it causes a much greater set of ebuilds to
- be cached.
-
-*portage-2.0.6 (25 Jun 2002)
-
- 25 Jun 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: rewrote the
- config file code to make things more sane, consistent, and made infinite
- loops during variable expansion a thing of the past. This closes bug #3952.
-
- 25 Jun 2002; Daniel Robbins <drobbins@gentoo.org> emerge, portage.py,
- ebuild.sh: "emerge --pretend" and "emerge search" now work for the root user
- as well as any users in the "wheel" group. emerge will now gracefully exit
- if "emerge search" is run by someone not in the "wheel" group. This closes
- bug #4121.
-
- 25 Jun 2002; Martin Schlemmer <azarah@gentoo.org> bin/{dolib.so,preplib.so}:
- Also change "strip --strip-unneeded" to "strip --strip-debug" for these.
- Look at bugs #2702,#3929,#4027 for more info.
-
- 25 Jun 2002; Daniel Robbins <drobbins@gentoo.org> emerge: "emerge -u" now
- correctly scans dependencies of packages, even if they are up-to-date, to
- determine if any of their dependencies need updating. Previously, up-to-date
- packages were not scanned in this way, which was a bug.
-
- 24 Jun 2002; Daniel Robbins <drobbins@gentoo.org> ebuild.sh: Portage will now
- automatically take advantage of ccache if the >=dev-util/ccache-1.9 ebuild is
- installed. Removing /var/cache/ccache hole in the sandbox.
-
-*portage-2.0.5 (24 Jun 2002)
-
- 23 Jun 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: basic keyword-
- based masking is now active. New support for KEYWORDS in ebuilds and
- ACCEPT_KEYWORDS in profiles.
-
- 23 Jun 2002; Daniel Robbins <drobbins@gentoo.org> conf files: removed
- references to non-functional FTP_PROXY and HTTP_PROXY and added a note about
- how to use the correct vars (ftp_proxy and http_proxy.) Closes bug #1664.
- Also added note about RSYNC_PROXY, closing bug #2332.
-
- 23 Jun 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: removed
- "dependency too short" checks which weren't working and were causing deps
- like "mc" and "ed" to be rejected as invalid rather than being expanded.
-
- 23 Jun 2002; Daniel Robbins <drobbins@gentoo.org> emerge: upgraded "clean"
- code to be more robust, consistent. Closes bug #3967.
-
-*portage-2.0.4 (20 Jun 2002)
-
- 20 Jun 2002; Daniel Robbins <drobbins@gentoo.org> in the 2.0.4 ebuild itself:
- added back tbz2tool symlink which was mistakenly removed. Also changed the
- way we compile portage.py stuff so that we remove the previously compiled
- files, just in case clock skew cause them to seem more recent than they
- really are. Python byte-code compilation moved to pkg_postinst()
-
- 20 Jun 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: fixed up the
- fetch() code so that invalid digests don't cause a traceback.
-
- 20 Jun 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: some of the new
- transplanted virtuals code assumed that profiledir was set, which is not
- required. This code has now been fixed to not make that assumption.
-
-*portage-2.0.3 (20 Jun 2002)
-
- 20 Jun 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: fixed two typos
- pointed out by stroke and g2boojum. These fix unmerge and some aspect of
- virtuals handling, which was previously causing a traceback.
-
- 19 Jun 2002; Daniel Robbins <drobbins@gentoo.org> emerge: emerge -eup world
- now will ignore world entries if it isn't really installed locally; in 2.0.1,
- --emptytree would auto-enable all world entries, leading to inconsistent
- package lists between --emptytree and without.
-
-*portage-2.0.2 (19 Jun 2002)
-
- 19 Jun 2002; Daniel Robbins <drobbins@gentoo.org> make.conf.ppc,
- make.globals.ppc: integrated these files into our sources.
-
- 19 Jun 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: fixed a bug
- where ROOT="" would not get properly converted to ROOT="/".
-
- 18 Jun 2002; Daniel Robbins <drobbins@gentoo.org> emerge: Portage 2.0+ has
- fixed "!" depend matching support, but some code in emerge was designed to
- anticipate the incorrect behavior. This emerge code has been fixed to work
- correctly. This closes my part of bug #3834.
-
- 18 Jun 2002; Daniel Robbins <drobbins@gentoo.org> portage.py, emerge: fixed
- bug #2444, where emerge fails when "--emptytree" and "--onlydeps" are used at
- the same time.
-
- 18 Jun 2002; Daniel Robbins <drobbins@gentoo.org> portage.py, emerge: new
- fakedbapi for doing emptytree calculations; emptytree upgrades in emerge.
- "emerge -upe world" now works correctly, accounting for items in the world
- profile. Yay! Bugs #3832 and #1911 fixed.
-
- 18 Jun 2002; Daniel Robbins <drobbins@gentoo.org> portage.py, emerge:
- Integrated new aux_get() method into our portdbapi. aux_get() provides a
- standardized way to get cached information about ebuilds, and is now fully
- integrated into Portage and emerge search.
-
- 18 Jun 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: fixed a bug
- where dep_nomatch() was testing for None rather than a "zero" condition,
- causing dep_nomatch() to choke on match()'s [] return value.
-
- 17 Jun 2002; Daniel Robbins <drobbins@gentoo.org> emerge: added an exception
- handler to catch invalid regular expressions and avoid a traceback.
-
- 17 Jun 2002; Daniel Robbins <drobbins@gentoo.org> sandbox.c: added
- /var/cache/ccache to sandbox "write ok" list, closing bug #3028.
-
-*portage-2.0.1 (16 Jun 2002)
-
- 16 Jun 2002; Daniel Robbins <drobbins@gentoo.org> portage.py, emerge: fixed a
- virtuals expansion bug that would cause virtuals to be consulted too early.
-
-*portage-2.0 (16 Jun 2002)
-
- 16 Jun 2002; Daniel Robbins <drobbins@gentoo.org> portage.py, ebuild.sh,
- emerge: turbo-ified Portage, new Portage db API and many other fixes.
- Unmerging now works again (bug introduced several releases ago.) I
- originally wanted to hold off on releasing 2.0, but since several major bugs
- are fixed, we need to get these fixes out to our users ASAP. We will be
- releasing 2.1 in about a week which should include additional refinements.
-
- 16 Jun 2002; Daniel Robbins <drobbins@gentoo.org> repoman: New repoman
- commit/check QA tool for developers.
-
- 10 Jun 2002; Dan Armak <danarmak@gentoo.org> make.conf: Update the ibiblio
- mirror path; it is now www.ibiblio.org/pub/Linux/distributions/gentoo.
-
- 10 Jun 2002; Martin Schlemmer <azarah@gentoo.org> ebuild.sh : Some users have
- $TMPDIR to a custom dir in their home ...this will cause sandbox errors with
- some ./configure scripts or libtool, so set it to $T.
-
- 10 Jun 2002; Martin Schlemmer <azarah@gentoo.org> portage.py : Merge in the
- ld.so.preload changes. Also updated spawn() to only run sandbox if
- buildphase is one of clean, unpack, compile or install. This should fix the
- handler not detecting some instances of sandbox running. Updated ebuildsh()
- to set buildphase="" on spawn exit.
-
- 10 Jun 2002; Martin Schlemmer <azarah@gentoo.org>
- portage.py.ldsopreload,testsandbox.sh : Add support to test if another
- sandbox is running, if so dont delete /etc/ld.so.preload on kill. I did not
- commit this to portage.py, as it is a bit more changes than we originally
- though.
-
- 05 Jun 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: doebuild() now
- has proper logic for digest generation. Digest will now get regenerated if
- "ebuild digest" is run, even if "digest" is in FEATURES.
-
- 04 Jun 2002; Daniel Robbins <drobbins@gentoo.org> emerge: fixed typo in
- "inject" documentation.
-
- 02 Jun 2002; Martin Schlemmer <azarah@gentoo.org> bin/dosym: changed the
- command used from "ln -sf" to "ln -snf" as it created a symlink in the target
- directory if the linkname already existed (only if the target is a
- directory). Im guessing this should be fixed in the python merged code if
- symlinks are not unlinked before the new is merged into place .. will add a
- bug later.
-
-*portage-1.9.14 (01 Jun 2002)
-
- 01 Jun 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: when a file to
- be merged is identical to the one currently on disk (using the md5 test), we
- now at least correctly update the target's mtime and atime so that cleaning
- will work correctly.
-
- 27 May 2002; Daniel Robbins <drobbins@gentoo.org> bin/do*: changed "return"s
- to "exit 1". Closes bug #3078.
-
-*portage-1.9.13 (21 May 2002)
-
- 21 May 2002; Grant Goodyear <g2boojum@gentoo.org> emerge.1: Updated man page.
- Thanks to carpaski@twobit.net.
-
- 20 May 2002; Daniel Robbins <drobbins@gentoo.org> emerge: a fix for dep
- regeneration; stale dep cache entries should now be properly regenerated.
-
-*portage-1.9.12 (16 May 2002)
-
- 16 May 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: the settings
- for A and AA were swapped; this has now been fixed, closing bug #1634.
-
- 15 May 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: fixed up some
- quirks in the new fetch code which were reported by Wout Mertens. Thanks
- Wout!
-
-*portage-1.9.11 (13 May 2002)
-
- 13 May 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: addition of
- third-party mirroring code. "mirror://sourceforge/foo.tar.gz" will use the
- /usr/portage/profiles/thirdpartymirrors file to define the mirror it will
- download from. Multiple mirrors for a single keyword can be specified on a
- single line. This code was based on the good work of Ryan Phillips.
-
- 13 May 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: (entry on behalf
- of jnelson) -- we now have new KV detection/setting code that uses
- /usr/src/linux/include/linux/version.h to determine the kernel version -- the
- right way of doing things. KV is set to "" if the kernel is not available or
- not configured.
-
- 13 May 2002; Daniel Robbins <drobbins@gentoo.org> emerge: "abspath[x]" =>
- "abspath(x)" typo fix. Also added 2-liner to allow for "emerge unmerge" to
- specify "foo/bar-1.0" rather than requiring "=foo/bar-1.0." Also fixed
- problems when specifying the names of actual ebuilds in /var/db/pkg to
- unmerge.
-
-*portage-1.9.9 (06 May 2002) ??
-
- 08 May 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: An empty
- USE_ORDER (due to an out-of-date /etc/make.globals) would cause USE to always
- be empty. I added a reasonable default USE_ORDER if USE_ORDER is not found
- in any of the config files, fixing this problem.
-
- 06 May 2002; Daniel Robbins <drobbins@gentoo.org> emerge: post bug #1841
- cleanups; converted from .hasnode() to .dep_match() (correct) in
- getworldlist().
-
-*portage-1.9.8 (06 May 2002)
-
- 06 May 2002; Daniel Robbins <drobbins@gentoo.org> emerge: 2 fixes that seemed
- to get fried/zapped: bug #1841 and fixing an emerge sync error code to be
- more understandable.
-
-*portage-1.9.7 (06 May 2002)
-
- 06 May 2002; Daniel Robbins <drobbins@gentoo.org> portage.py, emerge: initial
- bunch of robustness/error-handling fixes. Emerge should now report a
- comprehensible error message for errors in DEPEND and RDEPEND rather than
- giving a traceback. Fixed dep_depreduce() to catch errors rather than
- passing them on to dep_bestmatch() (which doesn't check for errors and
- assumes correct input)
-
- 02 May 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: typo fix for
- S_ISFIFO call; thanks woodchip!
-
-*portage-1.9.6 (02 May 2002)
-
- 02 May 2002; Daniel Robbins <drobbins@gentoo.org> make.conf: SYNC variable
- correctly set to use our DNS round-robin system (rsync.gentoo.org).
-
-*portage-1.9.6_pre2 (01 May 2002)
-
- 01 May 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: config file
- protection upgrade; we now record md5sums of previously-merged config protect
- files in /var/cache/edb/config; we use this information to avoid merging
- files that have been merged by us before (if it is safe to do so). This
- doesn't solve the problem of rolling back to a previously-merged version of
- a config file; we need to add cvs headers to every config file to get that
- to work.
-
-*portage-1.9.6_pre1 (01 May 2002)
-
- 01 May 2002; Daniel Robbins <drobbins@gentoo.org> emerge: added "inject"
- capability to artificially satisfy a dep -- for situations when you don't
- want Portage to do it for you because you've taken care of it already.
-
- 01 May 2002; Daniel Robbins <drobbins@gentoo.org> portage.py, emerge: totally
- rewrote emerge's unmerge code. The original code (added by Bevin) was a bit
- too cryptic and "big" for my taste. Also added two new capabilities to
- portage.py's vartree: .getslot() and .gettimeval(). getslot() returns the
- slot value of a cat/pkg-v, if any, and gettimeval() returns a "time value"
- (based on mtime/COUNTER value) that can be used to determine the order in
- which packages got merged. The purpose of these improvements were to revamp
- the existing Portage code that could only unmerge a db entry if there was a
- corresponding ebuild file. This conflicted with the ability to "inject"
- packages (see above), so it needed to be changed. Also fixed some bugs along
- the way.
-
- 01 May 2002; Daniel Robbins <drobbins@gentoo.org> ebuild.sh: einstall now
- accepts arguments, closing bug #2275. Thanks seemant! :)
-
- 01 May 2002; Daniel Robbins <drobbins@gentoo.org> make.globals: switch SYNC
- var from cvs.gentoo.org to rsync.gentoo.org, and gentoo-x86-portage to
- gentoo-portage.
-
- 30 Apr 2002; Daniel Robbins <drobbins@gentoo.org> emerge: now checks for
- errors (caused by bad deps) returned by create() and syscreate().
-
-*portage-1.9.5 (29 Apr 2002)
-
- 29 Apr 2002; Daniel Robbins <drobbins@gentoo.org> ebuild.sh: added jnelson's
- KV_extract.awk script which extracts the kernel version from the actual
- kernel sources Makefile, making our KV setting much more robust! Thanks Jon
- :)
-
- 29 Apr 2002; Daniel Robbins <drobbins@gentoo.org> emerge: correctly fixed
- emerge search examples to use single quotes (to turn off globbing.) This
- closes bug #1609.
-
- 29 Apr 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: virtual entries
- for no-longer-installed packages now get automatically removed on unmerge.
- This closes bug #2255 and #1891 (Thilo Bangert's comment on #1891 is also
- fixed.)
-
-*portage-1.9.4 (29 Apr 2002)
-
- 29 Apr 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: Added support
- for RESTRICT="nomirror". If "mirror" is defined in FEATURES and "nomirror"
- is defined in RESTRICT, then files will not be fetched.
-
- 29 Apr 2002; Daniel Robbins <drobbins@gentoo.org> emerge: Fixed docs for
- "unmerge", added docs for "--oneshot". Closes bugs #2156 and #2182.
-
- 29 Apr 2002; Daniel Robbins <drobbins@gentoo.org> ebuild.sh: Fix for USE
- troubles; confirmed by Azarah as working :) Also, we now strip the path
- from "KV" as we should.
-
-*portage-1.9.4_pre1 (26 Apr 2002)
-
- 26 Apr 2002; Daniel Robbins <drobbins@gentoo.org> tarball.sh: no longer
- necessary for VERSION to equal "@portage_version@" to get the version
- auto-set. VERSION can now be set to anything and tarball.sh will get it
- right.
-
- 26 Apr 2002; Daniel Robbins <drobbins@gentoo.org> emerge: I've started work
- on revamping/improving the "blocks" system to get ready to fix bug #1891.
- Already fixed a bug where "blocks" ("!" deps) print out the wrong blocking
- package name. This is now fixed. Blocks need additional testing.
-
- 26 Apr 2002; Daniel Robbins <drobbins@gentoo.org> portage.py, ebuild.sh, use:
- "use" is now a bash builtin function rather than an external python-based
- command. This appears to fix the USE inconstencies, as it should, thereby
- closing bug #2000.
-
-*portage-1.9.3 (24 Apr 2002)
-
- 24 Apr 2002; Jon Nelson <jnelson@gentoo.org>: portage.py, chkcontents:
- Emulate fchksum's md5 checksum routine. Closes bug #2787.
-
- 23 Apr 2002; Daniel Robbins <drobbins@gentoo.org>: emerge: invalid
- short options result in an exit. Closes bug #2025.
-
- 22 Apr 2002; Daniel Robbins <drobbins@gentoo.org>: emerge: rsync zlib
- compression enabled by default.
-
-*portage-1.9.2 (21 Apr 2002)
-
- 21 Apr 2002; Daniel Robbins <drobbins@gentoo.org>: portage.py: signal
- handler now zaps /etc/ld.so.preload if it's there -- this prevents it
- from hanging around and causing sandbox badness.
-
- 21 Apr 2002; Daniel Robbins <drobbins@gentoo.org>: emerge: fixed an
- error where --emptytree mode wouldn't really have a fully empty tree,
- due to the emptytree.inject() coming before the emptytree.root=None;
- the inject() caused a recalc of the USE vars, which caused the tree
- to become partially populated and it wasn't cleared. This closes bug
- #1897.
-
- 21 Apr 2002; Daniel Robbins <drobbins@gentoo.org>: portage.py: package
- building now executes the "setup" stage along with the unpack, compile,
- install stages so that global variables can be shared. This fixes an issue
- with woodchip's new apache ebuild and closes bug #1813.
-
-*portage-1.9.1 (16 Apr 2002)
-
- 16 Apr 2002; Daniel Robbins <drobbins@gentoo.org>: tiny (and forgotten)
- quick fix.
-
-*portage-1.9.0 (16 Apr 2002)
-
- 16 Apr 2002; Daniel Robbins <drobbins@gentoo.org>: portage.py: fetching
- should now try *all* alternate download locations, closing bug #1544.
- Yay!
-
- 15 Apr 2002; Daniel Robbins <drobbins@gentoo.org>: emerge: "emerge R" now
- works correctly, closing bug #1094.
-
- 15 Apr 2002; Daniel Robbins <drobbins@gentoo.org>: ebuild.sh: *'s and ?'s in
- DEPEND and SRC_URI syntax should no longer get glob-expanded to files in
- /usr/portage. This fixes some cryptic bugs. It also closes bug #1473.
-
- 15 Apr 2002; Daniel Robbins <drobbins@gentoo.org>: portage.py: emerge --debug
- now works again. Closes bug #1437.
-
- 15 Apr 2002; Daniel Robbins <drobbins@gentoo.org>: portage.py: moved argument
- check to the beginning of the doebuild() function to prevent "ebuild
- foo.ebuild fart" from causing the md5sums to be checked before recognizing
- that "fart" is not a valid command. Closes bug #1823.
-
- 15 Apr 2002; Daniel Robbins <drobbins@gentoo.org>: portage.py: 'USE="-foo"
- emerge bar' should now work correctly and consistently. Fixes to the config
- class. Closes bug #1455.
-
- 15 Apr 2002; Daniel Robbins <drobbins@gentoo.org>: portage.py: applied
- jnelson's patch to properly kill all children when interrupted with ^C
-
- 15 Apr 2002; Donny Davies <woodchip@gentoo.org>: make.conf: added a
- RESUMECOMMAND for lukemftp.
-
- 15 Apr 2002; Daniel Robbins <drobbins@gentoo.org>: portage.py: fixed up some
- "zing" logic inside the merge code (fixing variable being used before
- assignment errors)
-
- 15 Apr 2002; Jon Nelson <jnelson@gentoo.org>: src/sandbox/Makefile,
- src/sandbox/problems/Makefile, src/sandbox/problems/sandbox_dev_fd_foo.c,
- src/sandbox/sandbox.c: Cleaned up Makefiles somewhat to take advantage of GNU
- Make, added '/dev/zero' and /dev/fd/' (<- note trailing slash) to the list of
- items accessible safely from the sandbox. Added sandbox_dev_fd_foo.c to test
- for /dev/fd/<xx>. The test is almost verbatim from the autoconf test suite.
-
- 13 Apr 2002; Martin Schlemmer <azarah@gentoo.org>: ebuild.sh:
- Export $S and $D in dyn_compile and dyn_install, as our patched
- version of libtool uses these to fixup .la files.
-
-*portage-1.8.19 (09 Apr 2002)
-
- 08 Apr 2002; Geert Bevin <gbevin@gentoo.org>: emerge: Removed --all switch to
- emerge unmerge and clea. Added emerge prune which is the same as old emerge
- unmerge. Emerge unmerge removes all instances again without any proctedtion,
- as before. Added reporting of Omitted versions due to dep selectors.
- Renamed Removing and Keeping to Selected and Protected
-
- 08 Apr 2002; Geert Bevin <gbevin@gentoo.org>: emerge, portage.py: Better
- unmerge and clean reporting. Added homepage output to emerge --search.
- Bugfix when specifying a package without category that doesn't exist.
-
-*portage-1.8.18 (07 Apr 2002)
-
- 07 Apr 2002; Geert Bevin <gbevin@gentoo.org>: make.globals, emerge:
- Simplified unmerge functionality. Deprecated the CLEANMODE var,
- emerge clean now removes both revisions and slots automatically,
- emerge unmerge now removes all versions and revisions by default.
- Both understand world and system targets, and the --all option which doesn't
- check which packages are old and outdated.
- Documentation fixes.
- AUTOCLEAN var addition and added the autoclean functionality.
-
-*portage-1.8.17 (05 Apr 2002)
-
- 05 Apr 2002; Daniel Robbins <drobbins@gentoo.org>: portage.py: no longer
- print out the counter number after a merge. This information should not
- need to be known by end-users.
-
- 05 Apr 2002; Daniel Robbins <drobbins@gentoo.org>: sandbox.c: patched to
- allow access to /dev/vc from sandbox, allowing vim and screen to compile
- correctly from console.
-
- 05 Apr 2002; Daniel Robbins <drobbins@gentoo.org>: portage.py: fixed a bug
- that prevented md5 digests from being compared on archives.
-
- 04 Apr 2002; Daniel Robbins <drobbins@gentoo.org>: portage.py: Parse errors
- in /etc/env.d files no longer cause a traceback.
-
-*portage-1.8.16 (04 Apr 2002)
-
- 04 Apr 2002; Geert Bevin <gbevin@gentoo.org> portage.py, emerge : Forgot to
- commit my changes, this is an assembled version of both drobbins's and my
- changes
-
-*portage-1.8.15 (04 Apr 2002)
-
- 04 Apr 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: Fixed a
- semi-long standing bug where I was comparing atimes rather than mtimes when
- seeing if dep caches were stale. This fixes a bug originally reported by Dan
- Armak.
-
-*portage-1.8.14 (04 Apr 2002)
-
- 04 Apr 2002; Geert Bevin <gbevin@gentoo.org> emerge:
- Fixed short options for --all, --safe and clean.
-
-*portage-1.8.13 (04 Apr 2002)
-
- 04 Apr 2002; Geert Bevin <gbevin@gentoo.org> portage.py, emerge:
- Added emerge --all clean and emerge --safe unmerge.
- Made the counter updates atomic so that multiple merges can happen without
- risking counter clashes.
- Updated --help.
- Fixed short options bug that prevented two seperate short options to be
- specified successively.
-
-*portage-1.8.12 (04 Apr 2002)
-
- 04 Apr 2002; Geert Bevin <gbevin@gentoo.org> portage.py, emerge:
- implemented "emerge clean" with oldrevs, oldversions and oldslots options for
- the make.conf/make.globals CLEANMODE variable.
-
- 03 Apr 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: Added an
- indentation fix to prevent merge from dying if a file attempts to install
- itself on top of an existing directory. Closes bug #1498.
-
-*portage-1.8.11.1 (03 Apr 2002)
-
- 03 Apr 2002; Geert Bevin <gbevin@gentoo.org> portage.py: important fixes
- for movefile() -- changed the order of ownership / permissions settings since
- suid/guid bits were overwritten, added support back for 'ebuild config' since
- it was accidentally left out
-
-*portage-1.8.11 (02 Apr 2002)
-
- 02 Apr 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: important fixes
- for movefile() -- ownership now preserved across filesystems, mv -f fallback
- for special files, other optimizations and robustness improvements. Important
- fixes all-round.
-
- 02 Apr 2002; Geert Bevin <gbevin@gentoo.org> emerge:
- fixed bad indentation of a part of the code that made unmerging multiple
- packages behave badly
-
-*portage-1.8.10 (01 Apr 2002)
-
- 01 Apr 2002; Daniel Robbins <drobbins@gentoo.org> make.globals: removed USE
- settings from make.globals.
-
- 01 Apr 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: In trying to fix
- the glibc merge bug, we rewrote movefile() and it now runs blazingly fast. I
- also tweaked the code to remove the need for an ">>> Updating mtimes..."
- stage. This is all done dynamically now. *Much* faster. And hopefully
- solid for glibc upgraders.
-
-*portage-1.8.9.4 (01 Apr 2002)
-
- 31 Mar 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: adapted
- movefile() which should preserves timestamps and ownership on files and
- symlinks it moves.
-
-*portage-1.8.9.3 (31 Mar 2002)
-
- 31 Mar 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: promising
- rewrite of movefile() not only will probably fix glibc merge bug but also
- speeds up merging at least 20x! Calling "mv" for every file really make
- things super-slow!
-
-*portage-1.8.9.2 (31 Mar 2002)
-
- 31 Mar 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: desperate attempt
- to fix movefile() bug... a hack really. glibc merges still dying.
-
-*portage-1.8.9.1 (31 Mar 2002)
-
- 31 Mar 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: it appears that
- movefile() (using "/bin/mv") can still die if moving a new symlink on top of
- an existing library symlink. Upon failure, we now use a fallback mechanism
- to use "/bin/sln" to create the new symlink safely.
-
-*portage-1.8.9 (30 Mar 2002)
-
- 30 Mar 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: tiny cosmetic
- fix for digest generation.
-
- 30 Mar 2002; Daniel Robbins <drobbins@gentoo.org> emerge: entries in the
- world profile will be ignored if at least one version of the package in
- question isn't already merged. Prevents "--update world" from remerging
- packages that have since been unmerged.
-
- 30 Mar 2002; Daniel Robbins <drobbins@gentoo.org> various: changing all
- references from /usr/bin/python to /usr/bin/python2.2.
-
- 30 Mar 2002; Daniel Robbins <drobbins@gentoo.org> db-update.py: moving this
- script here from FILESDIR.
-
- 30 Mar 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: stuff doesn't
- get added to the world profile if it is already an essential ("*") package in
- the system profile. Keeps things clean and flexible.
-
-*portage-1.8.9_pre38 (30 Mar 2002)
-
- 30 Mar 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: resolve_key()
- didn't call load() on mapped virtuals, resulting in inconsistent resolution
- of virtual keys. Now fixed; thanks to woodchip for the bug report.
-
-*portage-1.8.9_pre37 (30 Mar 2002)
-
- 30 Mar 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: specific
- provides (dev-lang/python-2.2) will be converted to their generic form
- (dev-lang/python) before being recorded.
-
-*portage-1.8.9_pre36 (30 Mar 2002)
-
- 30 Mar 2002; Daniel Robbins <drobbins@gentoo.org> portage.py, emerge: new
- virtuals fixes and a db-upgrade.py script in the ebuild to solve another
- virtuals problem.
-
-*portage-1.8.9_pre35 (29 Mar 2002)
-
- 29 Mar 2002; Daniel Robbins <drobbins@gentoo.org> portage.py, emerge: digest
- generation fixes; emerge --update will now update dependents even if the main
- package hasn't been updated. getdict() fix, fixing a traceback.
-
-*portage-1.8.9_pre34 (28 Mar 2002)
-
- 28 Mar 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: Added an indent
- fix for the new virtuals code, eliminating a traceback. Tweaked digest handling,
- fixing a cosmetic error.
-
-*portage-1.8.9_pre33 (28 Mar 2002)
-
- 28 Mar 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: tweaked digest
- creation in doebuild() to hopefully avoid creating digests twice. I hope I
- didn't break anything in the process.
-
- 28 Mar 2002: Daniel Robbins <drobbins@gentoo.org> portage.py, emerge: "emerge
- update" is now "emerge --update"; "world" added back. New world
- implementation that works like this ... if you type "emerge media-gfx/gimp",
- media-gfx/gimp will be added to the "world" favorites file in
- /var/cache/edb/world. This world file is added to the system profile in
- order to create the set of "world" packages to merge. You can manually put
- specific deps (for pinning) in /var/cache/edb/world and they won't get
- overwritten. To prevent a package that you specify on the emerge
- command-line from being added to the world profile, use the "--oneshot"
- option, which tells emerge that it should be merged once but not updated.
-
- 28 Mar 2002: Daniel Robbins <drobbins@gentoo.org> portage.py: New virtuals
- implementation; new virtual info is recorded in /var/cache/edb/virtuals;
- virtuals data is created by merging the profile virtual info with the new
- edb/virtuals file. Note: We no longer add virtual package entries to
- /var/db/pkg. This change solves the bug where virtual files get auto-updated
- by pkg-update, and also solves the bug where doing an "emerge --update world"
- will cause ssmtp to be merged in order to satisfy the virtual/mta dep (which
- you already have satisfied by postfix, for example). Now, postfix will be
- updated if necessary, but that's it. :)
-
-*portage-1.8.9_pre32 (22 Mar 2002)
-
- 22 Mar 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: fixed a bug
- related to scanning for available binary packages, which would cause emerge
- to die.
-
- 21 Mar 2002; Dan Armak <danarmak@gentoo.org> portage.py: added ECLASSDIR
- (=$PORTDIR/eclass) to settings exported by python side to bash side.
- ebuild.sh: remove ECLASSDIR setting to use the one now provided by
- portage.py. Also, clean inherit() and make it use debug-print().
-
-*portage-1.8.9_pre31 (21 Mar 2002)
-
- 21 Mar 2002; Daniel Robbins <drobbins@gentoo.org> ebuild.sh: Added Dan
- Armak's patch to enable eclasses. :)
-
- 21 Mar 2002; Grant Goodyear <g2boojum@gentoo.org> portage.py:
- Added a drobbins patch to at line 469. Should fix emerge rsync
- problem when /etc/make.profile doesn't exist.
-
-*portage-1.8.9_pre30 (20 Mar 2002)
-
- 20 Mar 2002; Daniel Robbins <drobbins@gentoo.org> portage.py, emerge: emerge
- --emptytree now works again and I added a new --nodeps option to emerge as
- well.
-
-*portage-1.8.9_pre29 (20 Mar 2002)
-
- 20 Mar 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: CONTENTS files
- now record the correct path when ROOT!="/".
-
-*portage-1.8.9_pre28 (20 Mar 2002)
-
- 20 Mar 2002; Daniel Robbins <drobbins@gentoo.org> portage.py, emerge: an
- /etc/make.profile dir is now optional and portage will work if it's missing
- or is a broken symlink.
-
- 20 Mar 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: fix load()
- method in packagetree() class.
-
-*portage-1.8.9_pre27 (19 Mar 2002)
-
- 19 Mar 2002; Daniel Robbins <drobbins@gentoo.org> portage.py:
- /etc/make.profile/packages wasn't correctly masking core system packages;
- fixed.
-
-*portage-1.8.9_pre26 (18 Mar 2002)
-
- 18 Mar 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: configuration
- settings don't get loaded from ${ROOT}/etc anymore, just /etc. This
- simplifies the creation of a new build image.
-
- 18 Mar 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: Don't expect a
- digest file if SRC_URI is set to "" or undefined.
-
- 18 Mar 2002; Daniel Robbins <drobbins@gentoo.org>: RDEPEND was getting set
- to DEPEND even if there was an RDEPEND="" in the ebuild. This has now been
- fixed and RDEPEND will only get to DEPEND if RDEPEND has unset (not "unset
- or null", as it was before this fix.)
-
- 18 Mar 2002; Daniel Robbins <drobbins@gentoo.org>: emerge: circular
- dependencies error will result in the digraph dependencies being printed to
- stdout, greatly simplifying debugging.
-
- 18 Mar 2002; Geert Bevin <gbevin@gentoo.org>: src/sandbox/problems: Added
- several sample implementations to reproduce reported bugs.
-
- 18 Mar 2002; Geert Bevin <gbevin@gentoo.org>: libsandbox.c: All paths are now
- checked for multiple successive slashes anywhere, this closes bug 827.
- Performance should be slightly improved for other apps on the system for who
- the sandbox is not turned on.
-
- 14 Mar 2002; Daniel Robbins <drobbins@gentoo.org>: output.py: quick fix to
- turn off white color output -- not good for terminals with white background.
- This closes bug #1135.
-
- 11 Mar 2002; Daniel Robbins <drobbins@gentoo.org>: Applied karltk's version
- information patch. Portage.py now contains version information, and this
- info is reported by "emerge --version". Thanks Karl! :)
-
-*portage-1.8.9_pre25 (11 Mar 2002)
-
- 10 Mar 2002; Daniel Robbins <drobbins@gentoo.org> portage.py, emerge: roll
- back portagetree incremental db optimizations to fix an important bug (no
- package.mask calculations being done.) Emerge will now abort if a dependency
- can't be satisfied rather than printing just a warning.
-
-*portage-1.8.9_pre24 (09 Mar 2002)
-
- 09 Mar 2002; Daniel Robbins <drobbins@gentoo.org> emerge: fixed missing
- "--clean" option so "--clean rsync" should work again. Also converted emerge
- to work with our new db layout (db objects are in the portage module now).
-
- 09 Mar 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: Added USE
- ordering, a concept envisioned by Vitaly Kushneriuk. The idea is to allow
- customization of how Portage resolves USE variables. The default is
- "env:conf:auto:defaults" (set in the USE_ORDER variable -- this default is
- now in make.globals but as always you should set your custom USE_ORDER in
- make.conf if desired.) So, my default, USE settings will be calculated as
- follows. The USE variables will be grabbed from the
- /etc/make.profile/make.defaults file (this is "defaults".) Then, the
- /etc/make.profile/use.defaults file will be consulted. This file lists (one
- per line) a USE variable and a corresponding dependency. If the dependency
- is satisfied, the USE variable is auto-enabled and added to our working list
- (this is "auto"). Then /etc/make.conf is consulted, so that the user has the
- option of force-enabling (with USE="foo") or force-disabling (with
- USE="-foo") any USE variables as desired (this is "conf".) Then, the
- environment is consulted, allowing easy modifications from the command-line
- (this is "env"). Also added is the ability for USE settings to be
- dynamically regenerated in an efficient way as packages are merged. So,
- right after xfree is emerged, the "X" use variable can be immediately
- auto-enabled. Thanks to Vitaly for providing the vision for this.
-
-*portage-1.8.9_pre23 (08 Mar 2002)
-
- 09 Mar 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: our portage
- tree and /var/db/pkg tree now uses incremental caching rather than doing
- an exhaustive scan on startup. This should speed up this respective code
- by 2-30x depending on how many packages you're emerging. This involved
- quite a few changes to the code, and there could be some bugs, although I
- tried to be very careful. These changes could also break code that access
- internal .tree[] dictionaries directly.
-
- 08 Mar 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: SRC_URIs with
- USE-enabled ( ) clauses in them weren't working because I was iterating
- through multi-level lists without flatten()ing them. Now fixed.
-
- 07 Mar 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: We no longer
- depend upon a file being unmerged. We try to unlink during unmerge() in a
- "try" clause, with an exception for OSError. If the directory is immutable
- then Portage continues gracefully. This closes a bug reported by Verwilst --
- unmerging an old baselayout would die when it tried to delete /dev/shm.
- Fixed in Portage to eliminate these kinds of problems in the future.
-
-*portage-1.8.9_pre22 (07 Mar 2002)
-
- 07 Mar 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: I had
- sys.exit() in doebuild()'s fetch section. Converted to "return 1" and
- "return 0" to be emerge --fetchonly friendly. Thanks to Dan Armak for the
- bug report.
-
-*portage-1.8.9_pre21 (07 Mar 2002)
-
- 07 Mar 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: fixed symlink
- merge bug related to "safe" emerges. This should fix problems of Portage
- dying when merging certain things. I'm hoping that this fix will give us a
- solid merge implementation. It should. Thanks to Bart Verwilst for tracking
- down a merge problem that I could reproduce.
-
-*portage-1.8.9_pre20 (06 Mar 2002)
-
- 06 Mar 2002; Daniel Robbins <drobbins@gentoo.org> ebuild.sh, others: users
- can now forcefully disable stripping of their ebuild binaries by adding a
- RESTRICT="nostrip" to their ebuild. Additionally, the DEBUG variable has
- been renamed to DEBUGBUILD to prevent namespace collisions. This closes bug
- #868.
-
- 06 Mar 2002; Daniel Robbins <drobbins@gentoo.org> portage.py : Fixed a
- long-standing bug where the pkg_preinst() and pkg_postinst() functions being
- called came from the existing package ebuild in /var/db/pkg rather than the
- new ebuild being merged. pkg_prerm() and pkg_postrm() still use the ebuild
- in /var/db/pkg, which is correct. Thanks to Dan Armak for tracking down this
- bug.
-
-*portage-1.8.9_pre19 (05 Mar 2002)
-
- 05 Mar 2002; Daniel Robbins <drobbins@gentoo.org> portage.py : Fixed two bugs
- reported by Dan Armak relating to incomplete (not missing) message digests.
- They should now get regenerated automatically is "digest" is in FEATURES and
- not confuse the fetch code.
-
-*portage-1.8.9_pre18 (04 Mar 2002)
-
- 04 Mar 2002; Daniel Robbins <drobbins@gentoo.org> portage.py : Bug fix in
- dep_bestmatch(); rev comparison now works correctly and dep_bestmatch will
- properly handle 2-digit revisions in particular. This closes bug #952.
-
- 04 Mar 2002; Daniel Robbins <drobbins@gentoo.org> portage.py : Thanks to a
- bug report by Bruce Locke, the long-standing bug where symlinks with
- different mtimes would still get unmerged is now fixed. This allows glibc
- (of all things) to be unmerged safely. We needed this fix. Closes bug #964.
-
- 03 Mar 2002; Daniel Robbins <drobbins@gentoo.org> portage.py : KDEDIR ->
- KDEDIRS (fixing my typo)
-
-*portage-1.8.9_pre17 (27 Feb 2002)
-
- 27 Feb 2002; Daniel Robbins <drobbins@gentoo.org> emerge : --help
- documentation is now up-to-date; short options tweaked.
-
- 26 Feb 2002; Daniel Robbins <drobbins@gentoo.org> portage.py : Fixed an
- unmerge error caused by digests now being recorded in lowercase. unmerge now
- compensates for 100% backwards compatibility. Moved "continue" outside of
- "try" block since this is disallowed in Python. This fixes a python warning
- during fetch.
-
- 26 Feb 2002; Geert Bevin <gbevin@gentoo.org> bin/dohtml : Added support for
- installing .gif files too since quite some docs still ship with those instead
- of .jpg or .png.
-
-*portage-1.8.9_pre16 (25 Feb 2002)
-
- 25 Feb 2002; Daniel Robbins <drobbins@gentoo.org> emerge : emerge syntax
- changed back to "classic" style; replaced dots with spinner, did a major code
- cleanup and removed pieces that will be rewritten for 1.8.9 final. emerge
- documentation isn't up-to-date at all; I'm holding off on the --help rewrite
- until after 1.8.9 features have been finalized (soon!)
-
- 25 Feb 2002; Daniel Robbins <drobbins@gentoo.org> : Added RESTRICT variable
- to ebuilds. If RESTRICT="fetch" is set in the ebuild, it means that the
- files listed in SRC_URI are simply filenames and that the real files must be
- downloaded manually. This allows us to deal with realplayer, since
- overriding dyn_fetch is no longer an option now that we have the fetch code
- in python. Also added a cosmetic tweak to emerge during info file
- regeneration.
-
- 25 Feb 2002; Daniel Robbins <drobbins@gentoo.org> cnf/* : updated ${x} ->
- ${URI} in $FETCHCOMMAND.
-
- 24 Feb 2002; Daniel Robbins <drobbins@gentoo.org> Fixed another fetch bug
- where the download wouldn't cycle to alternate mirrors (it'd get stuck on
- the first download location)
-
-*portage-1.8.9_pre15 (24 Feb 2002)
-
- 24 Feb 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: some fixing of
- the merge code; added automatic digest generation when "digest" is in
- FEATURES; made the fetch() code not depend on a pre-existing digest. This
- should fix all known digest/fetch issues.
-
-*portage-1.8.9_pre14 (23 Feb 2002)
-
- 23 Feb 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: If you have a
- file that's hosted directly on our master ibiblio mirror, you can specify
- it in SRC_URI as "http://mirror/myfile.tar.bz2". "http://mirror" will be
- expanded to the first mirror defined in GENTOO_MIRRORS, or our master ibiblio
- mirror if no mirrors are defined. Groovy. This closes bug #627.
-
- 23 Feb 2002; Daniel Robbins <drobbins@gentoo.org> make.globals, emerge,
- portage.py: wget is no longer hardcoded; FETCHCOMMAND is enabled again, but
- uses ${FILE} and ${URI} instead of ${x} and ${y} now. Added download
- resuming, which requires the definition of RESUMECOMMAND (added to
- make.globals). Updated spawn() so that when it's called as spawn(foo,free=1)
- as an argument, sandboxing is turned off. Replaced all calls to
- os.system(foo) with spawn(foo,free=1), since os.system() messes with signal
- handling. Added a default SIGINT signal handler to portage.py so that ^C
- interrupts are handled correctly (portage will immediately exit with a return
- code of 1). This has been tested and works for ebuild and emerge. These
- additions should also close bug #407 and #760.
-
- 23 Feb 2002; Daniel Robbins <drobbins@gentoo.org> emerge: spython -> python
- fix; resolution of cvs merge conflict. Removed edepend assignment bug
- reintroduced in Geert's commit; added comment explaining why the new code is
- needed so it doesn't get removed again ;)
-
- 22 Feb 2002; Geert Bevin <gbevin@gentoo.org> emerge: removed spurious cvs
- conflict lines fixed bug in the cleanup code where different slots weren't
- handled too well
-
-*portage-1.8.9_pre13 (22 Feb 2002)
-
- 22 Feb 2002; Daniel Robbins <drobbins@gentoo.org> various: fix /bin/sh
- symlink merge problems and massively simplified movefile() code. Added back
- some emerge code so that "emerge --search rsync" doesn't run rsync.
-
-*portage 1.8.9_pre12 (22 Feb 2002)
-
- 22 Feb 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: everything
- should now generate *lowercase* md5sums, fixing problems with digest
- backwards compatibility.
-
- 22 Feb 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: changed
- position of digest generation call so that "ebuild digest" now works if the
- sources have not yet been downloaded.
-
-*portage 1.8.9_pre11 (22 Feb 2002)
-
- 22 Feb 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: unmerging now
- works correctly.
-
-*portage 1.8.9_pre10 (22 Feb 2002)
-
- 22 Feb 2002; Daniel Robbins <drobbins@gentoo.org>: "emerge" didn't like the
- new portage.py (fixed); A merge database bug (fixed). Fixing the merge bug
- should also result in merges happening much faster than before, maybe up to 3
- times as fast. We no longer resolve symlinks when testing protection paths.
-
-*portage 1.8.9_pre9 (21 Feb 2002)
-
- 21 Feb 2002; Daniel Robbins <drobbins@gentoo.org>: moved entirety of Portage
- download/digest code from ebuild.sh (bash) to portage.py (python). No
- support for custom FETCHCOMMANDs yet, but that's coming soon. Good news is
- that it appears to work well and allowed some cleanups and optimizations to
- doebuild(). SYNC support added. See cnf/make.conf for more info --
- basically, "emerge rsync" (now callable via "emerge sync" as well) supports a
- configurable "Portage server" that begins either with "rsync://" or "cvs://".
- "emerge sync" is now not only a clean front-end, but a configurable front-end
- to the Portage update process. I also removed dependency checking from
- 'ebuild'. This allowed me to removed some redundant code from portage.py,
- and seemed fine to do since 'ebuild' is more and more becoming a low-level
- developer tool.
-
- 21 Feb 2002; Daniel Robbins <drobbins@gentoo.org>: rewrite of cumulative USE
- setting code so that the new FEATURES (what used to be called MAINTAINER) is
- now cumulative and supports "-" and "-*" options. FEATURES is expanded using
- all config files, while USE ignores make.globals but uses everything else.
- Optimization of a couple parts of doebuild(). Upgraded expandpath() and used
- it in one place where it was removed accidentally (in the new merge code.)
- Master category list is now stored externally in
- ${PORTDIR}/profiles/categories for ease of maintenance.
-
- 21 Feb 2002; Daniel Robbins <drobbins@gentoo.org> emerge: now supports the
- EMERGE_OPTS make.conf variable for enabling emerge options by default.
- Closes bug #605.
-
- 21 Feb 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: dep_match() now
- works with * deps (again? Looks like the code got ripped out somehow); this
- closes bug #490.
-
- 21 Feb 2002; Daniel Robbins <drobbins@gentoo.org> global change from
- "spython" -> "python", now that Portage is slick enough to update shared
- libraries correctly.
-
- 21 Feb 2002; Grant Goodyear <g2boojum@gentoo.org> bin/chkcontents,
- man/chkcontents.1: New script to compare what's in a package's CONTENTS file
- with what's actually on the filesystem. Useful for discovering that a
- package "collision" has occurred.
-
- 21 Feb 2002; Daniel Robbins <drobbins@gentoo.org> ebuild.sh, portage.py:
- Dependencies should be no longer checked during the "unpack" stage. This
- closes bug #231. Added the $KV kernel version variable to ebuild.sh so that
- it's available for all ebuilds. This closes bug #599.
-
- 21 Feb 2002; Daniel Robbins <drobbins@gentoo.org> ebuild.sh: S now defaults
- to ${WORKDIR}/${P} if it isn't defined. That's right. Defining S is now
- optional :) This should eliminate around 1000 lines from our ports tree.
-
- 20 Feb 2002; Daniel Robbins <drobbins@gentoo.org> portage.py: Fixed some
- major bugs in the new merge/config protect code. Merging and config
- protection should now work.
-
- 18 Feb 2002; Daniel Robbins <drobbins@gentoo.org> portage.py, emerge,
- ebuild.sh : Rewrote merge code. Should merge symlinks only *after* the
- target has been merged (needs testing). Added lots of comments. Split
- dblink.merge() into 2 new functions -- .walktree() and .mergeme().
- Cleaned/optimized merging a good deal. Added special "-*" USE variable to
- unset *all* USE variables defined up until that point.
-
- 18 Feb 2002; Daniel Robbins <drobbins@gentoo.org> portage.py, emerge,
- ebuild.sh : Fixed USE bug (config.reset() threw away our cumulative USE),
- streamlined turbo dep calculations by moving edb cache entry creation to
- ebuild.sh. Started coding new dblink.merge() method that should be much
- cleaner, faster and merge symlinks safely.
-
- 17 Feb 2002; Daniel Robbins <drobbins@gentoo.org> portage.py : Added
- cumulative USE variable support. The final USE var is calculated
- cumulatively starting with make.profile, then make.conf, and then the
- environment. Any "-foo" option in USE will turn off a previously-defined USE
- setting. So, if you want to use the profile-default USE settings except turn
- "X" off, you add 'USE="-X"' to /etc/make.conf and that's it. This greatly
- simplifies USE variable maintenance since developers can now easily add new
- USE variables that default to 'on'. Simply add the USE variable and then add
- it to make.profile. It will then be included in everyone's USE variables
- automatically unless they explicitly "-newvar" in /etc/make.conf or in the
- local environment. This also enables easy one-shot disabling of USE
- variables. For example, to merge xchat without GNOME support (when "gnome"
- is in your /etc/make.conf USE variable), simply type "USE="-gnome" emerge
- net-irc/xchat". This is a lot easier than temporarily tweaking
- /etc/make.conf.
-
- 17 Feb 2002: Daniel Robbins <drobbins@gentoo.org> emerge : Added a quick hack
- to cache ebuild dependency info using extended attributes on XFS filesystems.
- Gives a factor of 10 speedup for dependency calculations. Will look into a
- generic caching solution that should offer similar performance increases on
- all filesystems. Note: the fact that this hack is on CVS means that emerge
- is currently in a hyper-experimental state and shouldn't be used right now.
-
- 16 Feb 2002; G.Bevin <gbevin@gentoo.org> emerge : added support for 'emerge
- cat/pkg-version' instead of always having to require 'emerge cat/pkg'
-
- 15 Feb 2002; G.Bevin <gbevin@gentoo.org> emerge : reimplemented the display
- of cleaned packages to clearly show all versions that are about to be removed
- and which versions are going to stay
-
- 13 Feb 2002; D.Robbins <drobbins@gentoo.org> emerge :
- added --emptytree option
-
- 13 Feb 2002; G.Bevin <gbevin@gentoo.org> emerge, portage.py :
- fixed little bug where emerge --clean rsync wasn't correctly handled anymore
- implemented all new functionalities of emerge --clean, this adds --slots,
- --versions and --all options together with world and system modes
-
- 12 Feb 2002; G.Bevin <gbevin@gentoo.org> emerge, portage.py :
- changed emerge --zap to emerge --clean
- removed parts of the already implemented slots functionality to be able to
- fall back to a more flexible implementation. Binary compatible slots will now
- mostly influence emerge during the --clean operation
-
- 12 Feb 2002; G.Bevin <gbevin@gentoo.org> emerge :
- added fallback check in case UNMERGE_DELAY hasn't been defined in
- make.globals or make.conf
-
-*portage 1.8.9_pre8 (18 Feb 2002)
-
- 12 Feb 2002; G.Bevin <gbevin@gentoo.org> emerge :
- changed the rebuild code to use depgraph instead, dramatically reducing code
- duplication
-
- 11 Feb 2002; G.Bevin <gbevin@gentoo.org> emerge :
- changed indentation of a code part that wasn't at the right level. It got the
- count of the non slot packages during --zap completely wrong
-
-*portage 1.8.9_pre7 (11 Feb 2002)
-
- 11 Feb 2002; G.Bevin <gbevin@gentoo.org> portage.py :
- fixed bug in packagetree.dep_match() where a ~ dependency is returned as a
- string instead of as a list
-
- 11 Feb 2002; G.Bevin <gbevin@gentoo.org> emerge :
- - removed support for --all and world in 'emerge --unmerge', by default now
- all matching packages are removed since with the new slots functionality
- old package versions should not exists anymore
- - colored package name red too when a blocking package is found during
- pretend operation
- - changed emerge invocation arguments to be in a new universal interface
- format this has been decided to be :
- 'emerge --action --option --option [packageset]'
- some features however don't respect this, but that's since they are not
- package installation related and are easier to use as straight commands
- eg: emerge rsync
- - added UNMERGE_DELAY var to make.globals and support it in emerge to obtain
- the number of seconds to wait
- - support for NOCOLOR="yes" as wel as NOCOLOR="true"
- - major speedups for emerge -search
- - added formatted package descriptions of the matches from emerge --search
- that nicely wrap at 80 chars
- - rewrote retrieval of package descriptions to support descriptions that are
- specified on multiple lines
- - added support for 'noslot' to --zap, --update and --rebuild
-
- 11 Feb 2002; G.Bevin <gbevin@gentoo.org> sandbox.c, libsandbox.c :
- added checks to see if the files where information is written to are really
- regular files and not symlinks
-
-*portage 1.8.9_pre6 (10 Feb 2002)
-
- 10 Feb 2002; G.Bevin <gbevin@gentoo.org> emerge :
- added support for 'emerge --unmerge world' which removes all old package
- versions from the system
- made 'emerge --unmerge' take binary compatibility slots into account, this
- prevents that packages with different versions but also different slots, are
- being unmerged
- added support for "--verbose" in "emerge rebuild" and "emerge world" to
- provide details about packages that aren't in the local tree anymore and to
- notify the user about which packages don't support slots
-
- 9 Feb 2002; G.Bevin <gbevin@gentoo.org> emerge :
- added support for the rebuild mode which rebuilds all the packages on your
- system for which a corresponding package could be found in to portage tree
-
- 9 Feb 2002; G.Bevin <gbevin@gentoo.org> portage.py :
- renamed getEbuildPaths() or vartree to getebuildpaths()
- fixed bug in the merging of binary packages
-
- 7 Feb 2002; G.Bevin <gbevin@gentoo.org> xpack.py :
- added additional argument to tbz2.getfile() which allows default content to
- be provided when the requested file couldn't be found in the tbz2 archive
-
- 7 Feb 2002; G.Bevin <gbevin@gentoo.org> ebuild.sh :
- added SLOT information as the third entry in the temporary deps file that is
- generated during ebuild depend,
-
- 7 Feb 2002; G.Bevin <gbevin@gentoo.org> portage.py :
- obtain SLOT information from the deps file instead of using the slotgrab()
- function, removed slotgrab() function, updated some comments
-
- 7 Feb 2002; G.Bevin <gbevin@gentoo.org> emerge :
- added SLOT information to the generated edepend var for binary packages
-
-*portage 1.8.9_pre5 (6 Feb 2002)
-
- 6 Feb 2002; G.Bevin <gbevin@gentoo.org> portage.py :
- fixed bugs where the old instance wasn't unmerged correctly if slots are
- identical, but ebuild version numbers not
-
- 6 Feb 2002; Grant Goodyear <g2boojum@gentoo.org> cnf/make.conf :
- Copied proxy lines from make.globals (bug 431).
-
- 6 Feb 2002; Vitaly Kushneriuk <vitaly@gentoo.org> portage.py :
- added missing SLOT param to merge(...).
-
-*portage 1.8.9_pre4 (6 Feb 2002)
-
- 6 Feb 2002; G.Bevin <gbevin@gentoo.org> portage.py :
- fixed some bugs in the handling of PROVIDES and virtual packages together
- with slots, they used code that wasn't upudated to the new dblink constructor
- virtuals now use "" as slot, resulting in normally to same behaviour as what
- has been done before.
-
- 5 Feb 2002; G.Bevin <gbevin@gentoo.org> emerge :
- sorted the packages to unmerge since this makes package names with revision
- endings appear before plain version numbers. This makes the unmerge code
- first remove the old and non slot aware packages before removing the newer
- alternative which is in fact exactly the same apart from the revision
- identifier.
-
-*portage 1.8.9_pre3 (5 Feb 2002)
-
- 5 Feb 2002; G.Bevin <gbevin@gentoo.org> portage.py :
- only let the backward compatibility algorithm kick in during unmerge since
- it's impossible to correctly detect is in general.
-
-*portage 1.8.9_pre2 (5 Feb 2002)
-
- 5 Feb 2002; G.Bevin <gbevin@gentoo.org> portage.py :
- fixed bug in the backward compatibility algorithm
-
-*portage 1.8.9_pre1 (5 Feb 2002)
-
- 4 Feb 2002; G.Bevin <gbevin@gentoo.org> ebuild.sh, emerge, portage.py :
- added binary compatibility slots, this also contains additional code to keep
- the unmerging of packages backwards compatible.
-
- 4 Feb 2002; G.Bevin <gbevin@gentoo.org> portage.py :
- added back support for the "ebuild help" command which seems to have
- disappeared somewhere along the updates
-
- 4 Feb 2002; G.Bevin <gbevin@gentoo.org> make.conf, make.globals :
- added description and default entry for the imlib USE variable
- added flag to prozilla to disable prozilla's waiting for a user's keypress
- when a failure occurs
-
- 4 Feb 2002; G.Bevin <gbevin@gentoo.org> output.py :
- bugfix, write read() function instead of red(), doh !
-
- 4 Feb 2002; G.Bevin <gbevin@gentoo.org> emerge, portage.py :
- implemented the ! dependency which prevents incompatible packages to be
- installed on the same system at the same time
-
- 4 Feb 2002; G.Bevin <gbevin@gentoo.org> emerge :
- most emerge invocation options now have alternative short flags
-
- 4 Feb 2002; G.Bevin <gbevin@gentoo.org> emerge :
- emerge now doesn't unmerge the last version of an installed version by
- default anymore, to really remove all instanced of packages the --all flag
- has to be used, the --safe flag is deprecated
-
- 3 Feb 2002; G.Bevin <gbevin@gentoo.org> output.py, emerge,make.globals :
- all output can now be turned to black and white by using the functions in
- output.py, this determines the mode by checking to NOCOLOR variable in
- make.conf or make.globals
-
- 3 Feb 2002; G.Bevin <gbevin@gentoo.org> emerge :
- made it possible to run emerge --help and --search as non root
-
-*portage 1.8.8-r1 (1 Feb 2002)
-
- 1 Feb 2002; G.Bevin <gbevin@gentoo.org> portage.py, ebuild.sh, ebuild.5 :
- added support for a pkg_setup() function which is executed before anything
- else and can be typically used for package configuration actions or required
- system checks
-
- 1 Feb 2002; G.Bevin <gbevin@gentoo.org> portage.py :
- implemented the noauto MAINTAINER flags for all relevant ebuild commands
-
- 1 Feb 2002; G.Bevin <gbevin@gentoo.org> portage.py :
- make env-update disregard backup files
-
- 1 Feb 2002; G.Bevin <gbevin@gentoo.org> emerge :
- added --pretend support instead of interactively asking to proceed,
- also added a delay before unmerging though to be sure
-
-*portage 1.8.8 (1 Feb 2002)
-
- 1 Feb 2002; G.Bevin <gbevin@gentoo.org> emerge :
- added --safe switch to complement the --unmerge option
-
- 1 Feb 2002; G.Bevin <gbevin@gentoo.org> portage.py :
- rewrote the packagename without category support to also graceously handle
- deps specifiers at the beginning, this shouldn't be used in the ebuilds, but
- are very handy when using emerge --unmerge
-
- 1 Feb 2002; G.Bevin <gbevin@gentoo.org> emerge :
- added emerge --unmerge support
-
- 1 Feb 2002; G.Bevin <gbevin@gentoo.org> portage.py :
- added packagename without category support in the dep_match function
-
-*portage 1.8.7 (30 Jan 2002)
-
- 30 Jan 2002; G.Bevin <gbevin@gentoo.org> portage.py :
- integrated and slightly adapted Brent Rahn's code to support package names
- without category
-
- 30 Jan 2002; G.Bevin <gbevin@gentoo.org> ebuild.sh :
- integrated and fixed Azarah's patch to fix the wrongly generated archive size
- in the digests
-
-*portage 1.8.6-r3 (28 Jan 2002)
-
- 28 Jan 2002; G.Bevin <gbevin@gentoo.org> portage.py :
- disabled warnings about non existant config file paths
-
- 28 Jan 2002; G.Bevin <gbevin@gentoo.org> emerge :
- added verwilst pkgsearch code, which was turned into a seperate class and
- refactored for clarity and execution speed
-
- 28 Jan 2002; G.Bevin <gbevin@gentoo.org> portage.py :
- renamed _xxx vars to _prepart in Vitaly's code addition
-
-*portage 1.8.6-r2 (27 Jan 2002)
-
- 24 Jan 2002; Vitaly Kushneriuk <vitaly@gentoo.org> portage.py :
- Fixed version compare code Also added test script to test future versions.
-
-*portage 1.8.6-r1 (24 Jan 2002)
-
- 24 Jan 2002; Karl Trygve Kalleberg <karltk@gentoo.org> dojar:
- Fixed typos.
-
-*portage 1.8.6 (23 Jan 2002)
-
- 22 Jan 2002; G.Bevin <gbevin@gentoo.org> libsandbox.c :
- added an additional check for SANDBOX_ON to optimize the speed in the execvp
- function call. Also removed error messages being printed when the PATH var
- isn't set.
-
- 20 Jan 2002; Karl Trygve Kalleberg <karltk@gentoo.org> dojar :
- added dojar shell command as a java JAR handler
-
- 17 Jan 2002; Daniel Robbins <drobbins@gentoo.org> :
- The package chosen by "emerge sys-apps/shadow" now matches that chosen in an
- emerge update or emerge system. I forgot to add some "*" dep code to the
- dep_nomatch() method; this is now fixed.
-
-*portage 1.8.5 (13 Jan 2002)
-
- 13 Jan 2002; G.Bevin <gbevin@gentoo.org> ebuilds.sh, portage.py:
- added fine grained maintainer settings
-
-*portage 1.8.4 (13 Jan 2002)
-
- 12 Jan 2002; Daniel Robbins <drobbins@gentoo.org> :
- "emerge sys-apps/bash/" now works. (trailing "/" stripped to make
- tab-completion users happy) This fixes bug #119
-
- 12 Jan 2002; Daniel Robbins <drobbins@gentoo.org> :
- Portage should no longer bomb out if the current working directory doesn't
- exist (has been deleted from underneath).
-
- 12 Jan 2002; Daniel Robbins <drobbins@gentoo.org> :
- added "--world" option for "emerge update". This tells Portage to update
- the base system *as well as* upgrade any packages that are currently
- installed but have new versions available. This is your standard full system
- update command. This fixes bug #122
-
- 12 Jan 2002; Daniel Robbins <drobbins@gentoo.org> :
- Added an expandpath() cache which speeds up merging dramatically.
-
-*portage 1.8.3 (11 Jan 2002)
-
- 11 Jan 2002; Mikael Hallendal <hallski@gentoo.org> ebuild.sh:
- added functions econf and einstall. Also made src_compile having a default
- implementation.
-
-*portage 1.8.2 (07 Jan 2002)
-
- 07 Jan 2002; G.Bevin <gbevin@gentoo.org> :
- sandbox included in portage
-
-*portage 1.8.1 (30 Dec 2001)
-
- 29 Dec 2001; Daniel Robbins <drobbins@gentoo.org> :
- "emerge update" is now functional! Although the algorithm really should be
- sound, it is still considered in testing since we may need to tweak some
- dependencies.
-
- 27 Dec 2001; Daniel Robbins <drobbins@gentoo.org> :
- Removed objprelink from the default USE variable.
-
-*portage 1.8.0 (22 Dec 2001)
-
- 22 Dec 2001; Daniel Robbins <drobbins@gentoo.org> :
- I messed up the "transparent .so library" update. Included is the correct
- fix, which seems to allow glibc to update on my system without any tricks.
- I just merged a new glibc while in X, without using any of glibc's old
- pkg_postinst/pkg_preinst hacks.
-
- 21 Dec 2001; Daniel Robbins <drobbins@gentoo.org> make.defaults.5, make.conf.5 :
- Moved make.defaults.5 to make.conf.5 and updated contents.
-
-*portage 1.7.8 (21 Dec 2001)
-
- 21 Dec 2001; Daniel Robbins <drobbins@gentoo.org> portage.py :
- ebuild install doesn't throw away the compile directory if ebuild has been
- changed and MAINTAINER is set; we now use mv -f to move files into location
- on the filesystem, making our library install method even more robust. It
- should now be totally safe to upgrade glibc while in X and playing music
- with xmms and surfing the Web and compiling 20 applications at once :)
-
-* portage 1.7.7 (14 Dec 2001)
-
- 14 Dec 2001; Aron Griffis <agriffis@gentoo.org>: portage.py :
- movefile() now unlink()s the destination file first which solves shared
- library install problems.
-
-* portage 1.7.6 (13 Dec 2001)
-
- 10 Dec 2001; Daniel Robbins <drobbins@gentoo.org> portage.py :
- env_update() now generates an /etc/csh.env file in csh shell format.
-
-* portage 1.7.5 (13 Dec 2001)
-
- 10 Dec 2001; Daniel Robbins <drobbins@gentoo.org> portage.py :
- added a little fix so a ~ dep isn't satisfied by a *much* later installed
- version; i.e. ~media-libs/freetype-1.3.1 satisfied by
- media-libs/freetype-2.0.5 being installed.
-
- 30 Nov 2001; Daniel Robbins <drobbins@gentoo.org> ebuild :
- added a KeyboardInterrupt handler so ^C'ing a running ebuild process doesn't
- produce a Python traceback.
-
-* portage 1.7.4 (29 Nov 2001)
-
- 29 Nov 2001; Daniel Robbins <drobbins@gentoo.org> portage.py :
- removed an unnecessary quirk in the code preventing virtual packages from
- being in any other category but "virtual".
-
- 25 Nov 2001; Daniel Robbins <drobbins@gentoo.org> emerge :
- info file detection code has been improved.
-
-* portage 1.7.3 (20 Nov 2001)
-
- 20 Nov 2001; Daniel Robbins <drobbins@gentoo.org> pdb, pdb.cgi, xpak :
- used /usr/bin/python instead of /usr/bin/spython. Fixed.
-
-* portage 1.7.2 (13 Nov 2001)
-
- 13 Nov 2001; Aron Griffis <agriffis@gentoo.org> :
- emake will now try to build in parallel, and if it fails, will retry in
- non-parallel mode.
-
- 13 Nov 2001; Daniel Robbins <drobbins@gentoo.org> ebuild.sh :
- an ebuild that used an archive with a name that was part of another archive
- would cause Portage to mess up the digest check. No longer.
-
- 31 Oct 2001; Donny Davies <woodchip@gentoo.org> make.conf, make.globals :
- comments about merging the fetch programs before trying to use them.
-
-* portage 1.7.1 (30 Oct 2001)
-
- 30 Oct 2001; Daniel Robbins <drobbins@gentoo.org> portage.py :
- I accidentally broke "=" deps by being sloppy when I added "*" deps. Trivial
- fix applied.
-
-* portage 1.7 (29 Oct 2001)
-
- 29 Oct 2001; Daniel Robbins <drobbins@gentoo.org> portage.py :
- strange typo fix in the empty() digraph method.
-
-* portage 1.6.12 (29 Oct 2001)
-
- 29 Oct 2001; Daniel Robbins <drobbins@gentoo.org> portage.py :
- "*" deps now implemented. "=sys/foo-1*" matches the latest 1.x version/rev of
- sys/foo, but will not match a 2.x version.
-
- 26 Oct 2001; Daniel Robbins <drobbins@gentoo.org> portage.py :
- various new categories added.
-
-Portage 1.6.11, released 18 Oct 2001
-====================================
-
- *portage.py; device nodes are now not unmerged at all.
-
-
-Portage 1.6.10, released 18 Oct 2001
-===================================
-
- *ebuild.sh: A and AA fix; now any A="foo" lines in the ebuild are ignored,
- allowing ebuilds with them to continue working.
-
- *portage.py: remove gnome-apps, gnome-office, gnome-libs, add gnome-extra
- categories.
-
- *ebuild.sh, portage.py: You can now use USE variables in SRC_URI to
- conditionally include archives. In MAINTAINER mode, all archives are
- automatically included so that maintainers can check SRC_URIs and also
- generate complete digests. A new file-based DEPEND and RDEPEND-passing
- mechanism has been added.
-
- *ebuild.sh Now adds filesize to the digest files
-
-Portage 1.6.9, released 15 Sep 2001
-===================================
-
- *portage.py
- unmerge() now does not touch device nodes. Unlinking them or touching them
- in any way is bad practice.
-
-
-Portage 1.6.8, released 12 Sep 2001
-===================================
-
- *portage.py
- movefile() will now unlink() destfile if it is a symlink. Should fix
- problems where file gets created at symlink target rather than replacing
- the symlink.
-
- *queryhost.sh (agriffis)
- Parallel pinging and other fixes.
-
-Portage 1.6.7, released 05 Sep 2001
-===================================
-
- *portage.py
- (last-minute fix)
- CONFIG_PROTECT unmerge protection is now observed for symlinks, fifos and
- device nodes.
-
- *portage.py
- Fixes for unmerging CONFIG_PROTECTed files. Protected files are now *not*
- unlinked from the filesystem. Messier but safer, and simplifies package
- upgrades.
-
- *emerge
- Emerge output cleanups for GNU info directory generation. New --verbose
- mode; new CONFIG_PROTECT scanning feature to let people know when there are
- config files to be updated. New "--help config" docs to explain how to
- do it.
-
-Portage 1.6.6, released 01 Sep 2001
-====================================
-
- *portage.py
- Upgraded directory merging over existing objects. Existing symlinks that
- point to existing directories will be kept and used as is; directories will
- be used as-is; any other objects (broken symlinks, files) will be copied
- to origfilename.backup and a ne directory will be created in its place.
-
-Portage 1.6.5, released 31 Aug 2001
-====================================
-
- *emerge
- Fixes for handling multiple ebuilds, packages and/or dependencies on
- the command-line.
-
- *portage.py
- An optimization to the digraph class so that the firstzero() method finds
- matches in close to the order that keys were added to the digraph.
foo-cvs.1.ebuild in the same directory. cvs ebuilds will always be
ranked higher than any non-cvs ebuild.
- multiple suffixes now allow for constructions like foo-1.0_pre1_rc5
- Also the new code is now separated into its own module portage_versions.py
- with no dependencies on the main portage module. [genone]
+ [genone]
* recursive config files: Most of the config files in /etc/portage can now
also be directories, if they are then portage will recursively use all
files in them. So you could have /etc/portage/package.keywords/kde/base
and /etc/portage/package.keywords/gnome instead of a single package.keywords.
- This applies to the following files:
- - package.keywords
- - package.mask
- - package.unmask
- - package.use
- - mirrors
[genone]
+* multi-hash support: digests and Manifests will now use SHA256 and RMD160 in
+ addition to MD5 [genone]
Bug fixes / Minor Changes
-------------------------
* Removal of "masked by -*" message [genone]
* Fixed some minor screwup in the sort algorithm for versions [genone]
+* Make digest errors more verbose (list actual and expected checksums)
except KeyboardInterrupt:
print "Interrupted."
a = 1
+ except KeyError:
+ # aux_get error
+ a = 1
if a == None:
print "Could not run the required binary?"
a = 127
export SANDBOX_PREDICT="${SANDBOX_PREDICT}:/proc/self/maps:/dev/console:@PORTAGE_BASE@/pym:/dev/random"
-export SANDBOX_WRITE="${SANDBOX_WRITE}:/dev/shm:${PORTAGE_TMPDIR}"
-export SANDBOX_READ="${SANDBOX_READ}:/dev/shm:${PORTAGE_TMPDIR}"
+export SANDBOX_WRITE="${SANDBOX_WRITE}:/dev/shm:/dev/stdout:/dev/stderr:${PORTAGE_TMPDIR}"
+export SANDBOX_READ="${SANDBOX_READ}:/dev/shm:/dev/stdin:${PORTAGE_TMPDIR}"
if [ ! -z "${PORTAGE_GPG_DIR}" ]; then
SANDBOX_PREDICT="${SANDBOX_PREDICT}:${PORTAGE_GPG_DIR}"
return 0
}
+add-ebuild-death-hook() {
+ export EBUILD_DEATH_HOOKS="${EBUILD_DEATH_HOOKS} $*"
+}
+
diefunc() {
local funcname="$1" lineno="$2" exitcode="$3"
shift 3
if [ -s config.log ]; then
echo
- echo "!!! Please attach the config.log to your bug report:"
+ echo "!!! Please attach the following file when filing a report to bugs.gentoo.org:"
echo "!!! ${PWD}/config.log"
fi
die "econf failed"
if [[ -n ${f} ]] ; then
echo -ne '\a\n'
echo "QA Notice: the following files contain runtime text relocations"
- echo " Text relocations require a lot of extra work to be preformed by the"
- echo " dynamic linker which will cause serious performance impact on IA-32"
- echo " and might not function properly on other architectures hppa for example."
- echo " If you are a programmer please take a closer look at this package and"
- echo " consider writing a patch which addresses this problem."
+ echo " Text relocations force the dynamic linker to perform extra"
+ echo " work at startup, waste system resources, and may pose a security"
+ echo " risk. On some architectures, the code may not even function"
+ echo " properly, if at all."
echo "${f}"
echo -ne '\a\n'
qa_kinda_sucks=1
def __init__(self):
self.invalid = []
self.not_installed = []
+ self.masked = []
self.unavailable = []
self.okay = []
self.found = os.access(portage_const.WORLD_FILE, os.R_OK)
self.invalid.append(atom)
elif not portage.db["/"]["vartree"].dbapi.match(atom):
self.not_installed.append(atom)
- elif not portage.db["/"]["porttree"].dbapi.match(atom):
+ elif not portage.db["/"]["porttree"].dbapi.xmatch("match-all", atom):
self.unavailable.append(atom)
+ elif not portage.db["/"]["porttree"].dbapi.match("match-visible", atom):
+ self.masked.append(atom)
else:
self.okay.append(atom)
if self.found:
errors += map(lambda x: "'%s' is not a valid atom" % x, self.invalid)
errors += map(lambda x: "'%s' is not installed" % x, self.not_installed)
+ errors += map(lambda x: "'%s' is masked (manually fix by adding to package.keywords)" % x, self.masked)
errors += map(lambda x: "'%s' has no ebuilds available" % x, self.unavailable)
+ if self.masked:
+ errors += "Masked ebuilds can be manually unmasked with package.keywords or package.unmask"
else:
errors.append(portage_const.WORLD_FILE + " could not be opened for reading")
return errors
if ("getbinpkg" in portage.features):
myopts.append("--getbinpkg")
+if "--skipfirst" in myopts and "--resume" not in myopts:
+ myopts.append("--resume")
+
if ("--getbinpkgonly" in myopts) and not ("--usepkgonly" in myopts):
myopts.append("--usepkgonly")
elif portage.secpass!=2:
if myaction in ["search", "info", "regen", "metadata"]:
pass
- elif (not myaction) and (not myfiles):
+ elif not myaction and not myfiles and "--resume" not in myopts:
pass
elif ("--pretend" in myopts) and (myaction in ["world","system","clean","prune","unmerge"]):
pass
if curslot == myslot:
return match
- def select_dep(self,myroot,depstring,myparent=None,arg=None,myuse=None):
+ def select_dep(self,myroot,depstring,myparent=None,arg=None,myuse=None,raise_on_missing=False):
"given a dependency string, create the appropriate depgraph and return 1 on success and 0 on failure"
if "--debug" in myopts:
print
print "Depstring:",depstring
if not arg:
#processing dependencies
- mycheck=portage.dep_check(depstring,self.mydbapi[myroot],self.pkgsettings,myuse=myuse,use_binaries=("--usepkg" in myopts))
- #mycheck=portage.dep_check(depstring,self.mydbapi[myroot],self.pkgsettings,myuse=myuse)
+ mycheck=portage.dep_check(depstring,self.mydbapi[myroot],self.pkgsettings,myuse=myuse,use_binaries=("--usepkgonly" in myopts),myroot=myroot)
if not mycheck[0]:
mymerge=[]
break
if (not myeb) and (not myeb_pkg):
+ if raise_on_missing:
+ raise ValueError
if not arg:
xinfo='"'+x+'"'
else:
def xcreate(self,mode="system"):
global syslist
+ world_problems = False
if mode=="system":
mylist=syslist
else:
sysdict=genericdict(syslist)
worlddict=genericdict(worldlist)
- world_problems = False
for x in worlddict.keys():
if not portage.isvalidatom(x):
world_problems = True
elif not portage.db["/"]["vartree"].dbapi.match(x):
world_problems = True
- elif not portage.db["/"]["porttree"].dbapi.match(x):
- world_problems = True
else:
sysdict[x]=worlddict[x]
- if world_problems:
- print "\n!!! Problems have been detected with your world file"
- print "!!! Please run "+green("emaint --check world")+"\n"
- del world_problems
mylist = sysdict.keys()
+ newlist = []
+ for atom in mylist:
+ if portage.dep_getkey(atom).split("/")[-1] == "portage":
+ newlist.insert(0, atom)
+ else:
+ newlist.append(atom)
+ mylist = newlist
+
for mydep in mylist:
- if not self.select_dep(portage.root, mydep):
- print "\n\n!!! Problem resolving dependencies for", mydep
- return 0
+ try:
+ if not self.select_dep(portage.root, mydep, raise_on_missing=True):
+ print "\n\n!!! Problem resolving dependencies for", mydep
+ return 0
+ except ValueError:
+ if "--debug" in myopts:
+ raise
+ world_problems = True
+
+ if world_problems:
+ print "\n!!! Problems have been detected with your world file"
+ print "!!! Please run "+green("emaint --check world")+"\n"
+
return 1
def match(self,mydep,myroot=portage.root,mykey=None):
totalsize=0
if "--verbose" in myopts:
- overlays = string.split(portage.settings['PORTDIR_OVERLAY'])
- use_expand = portage.settings["USE_EXPAND"].lower().split()
- use_expand_hidden = portage.settings["USE_EXPAND_HIDDEN"].lower().split()
-
- def create_use_string(iuse, cur_use, old_use, masked_use):
- usestr=""
- for flag in iuse:
- usechange=""
- if old_use:
- if (flag in old_use and flag not in cur_use) or (flag not in old_use and flag in cur_use):
- usechange="*"
-
+ # Full USE display
+ def create_use_string(name, cur_iuse, cur_use, old_iuse, old_use, is_new):
+ usestrs = []
+ for flag in cur_iuse:
+ suffix = ""
if flag in cur_use:
- if usechange == "*":
- substr = green("+"+flag)
- else:
- substr = red("+"+flag)
- elif flag in masked_use:
- substr = blue("(-"+flag+")")
+ color = red
+ prefix = ""
else:
- substr = blue("-"+flag)
- usestr += substr + usechange + " "
- return usestr
+ color = blue
+ prefix = "-"
+ if not is_new and ((flag in old_use) ^ (flag in cur_use)):
+ color = green
+ suffix = "*"
+ usestrs.append(color(prefix+flag)+suffix)
+ ret = " ".join(usestrs)
+ if ret:
+ ret = '%s="%s" ' % (name, ret)
+ return ret
+ else:
+ # Changed-since-previous USE display
+ def create_use_string(name, cur_iuse, cur_use, old_iuse, old_use, is_new):
+ enabled = []
+ disabled = []
+ for flag in cur_iuse:
+ if flag in cur_use and flag not in old_use:
+ if flag in old_iuse:
+ enabled.append(green(flag)+"*")
+ else:
+ enabled.append(red(flag))
+ elif flag not in cur_use and flag in old_use:
+ disabled.append(green("-"+flag)+"*")
+ elif is_new or flag not in old_iuse:
+ disabled.append(blue("-"+flag))
+ enabled = " ".join(enabled)
+ disabled = " ".join(disabled)
+ if enabled and disabled:
+ ret = enabled + " " + disabled
+ elif enabled:
+ ret = enabled
+ else:
+ ret = disabled
+ if ret:
+ ret = '%s="%s" ' % (name, ret)
+ return ret
+
+ if "--verbose" in myopts:
+ overlays = string.split(portage.settings['PORTDIR_OVERLAY'])
if "--tree" in myopts:
mylist.reverse()
addl=" "+green("N")+" "+fetch+" "
verboseadd=""
- if "--verbose" in myopts:
- # iuse verbose
- try:
- if x[0] == "binary":
- iuse_split = string.split(portage.db["/"]["bintree"].dbapi.aux_get(x[2],["IUSE"])[0])
- elif x[0] == "ebuild":
- iuse_split = string.split(portage.portdb.aux_get(x[2],["IUSE"])[0])
- else:
- iuse_split = []
- except SystemExit, e:
- raise # Needed else can't exit
- except:
- portage.writemsg("!!! Error getting IUSE (report this to bugs.gentoo.org)\n")
- portage.writemsg("!!! %s\n" % x)
- iuse_split = []
-
- iuse_split = portage.unique_array(iuse_split)
- iuse_split.sort()
+
+ if x[2] in self.applied_useflags:
+ # USE flag display
+ if x[0] == "binary":
+ cur_iuse = string.split(portage.db["/"]["bintree"].dbapi.aux_get(x[2],["IUSE"])[0])
+ elif x[0] == "ebuild":
+ cur_iuse = string.split(portage.portdb.aux_get(x[2],["IUSE"])[0])
+ else:
+ cur_iuse = []
+ cur_iuse = portage.unique_array(cur_iuse)
+ cur_iuse = [flag for flag in cur_iuse if flag not in portage.settings.usemask]
+ cur_iuse.sort()
cur_use = self.applied_useflags[x[2]]
+ cur_use = [flag for flag in cur_use if flag in cur_iuse]
- old_use = []
if myoldbest:
- pkg=myoldbest
+ pkg = myoldbest
else:
- pkg=x[2]
+ pkg = x[2]
if portage.db["/"]["vartree"].dbapi.cpv_exists(pkg):
- try:
- old_use=string.split(portage.db["/"]["vartree"].dbapi.aux_get(pkg, ["USE"])[0])
- except SystemExit, e:
- raise # Needed else can't exit
- except:
- pass
-
- reg_use = []
- exp_map = {}
- for flag in iuse_split:
- found = False
- for var in use_expand:
- if flag.startswith(var+"_"):
- if var in exp_map:
- exp_map[var]+= [flag[len(var)+1:]]
- else:
- exp_map[var] = [flag[len(var)+1:]]
- found = True
- break
- if not found:
- reg_use.append(flag)
-
- usestr = create_use_string(reg_use, cur_use, old_use, portage.settings.usemask).strip()
- if usestr:
- verboseadd += 'USE="'+usestr+"' "
-
- for var in use_expand:
- if var not in exp_map:
- continue
- expcur = []
- expold = []
- expmask = []
- for flag in exp_map[var]:
- if var+"_"+flag in cur_use:
- expcur+= [flag]
- if var+"_"+flag in old_use:
- expold+= [flag]
- if var+"_"+flag in portage.settings.usemask:
- expmask+= [flag]
- verboseadd += var.upper()+'="'+create_use_string(exp_map[var],expcur,expold,expmask).strip()+'" '
+ (old_iuse, old_use) = portage.db["/"]["vartree"].dbapi.aux_get(pkg, ["IUSE", "USE"])
+ old_iuse = portage.unique_array(old_iuse.split())
+ old_iuse.sort()
+ old_use = old_use.split()
+ is_new = False
+ else:
+ old_iuse = []
+ old_use = []
+ is_new = True
+ old_iuse = [flag for flag in old_iuse if flag not in portage.settings.usemask]
+ old_use = [flag for flag in old_use if flag in old_iuse]
+
+ use_expand = portage.settings["USE_EXPAND"].lower().split()
+ use_expand.sort()
+ use_expand.reverse()
+ use_expand_hidden = portage.settings["USE_EXPAND_HIDDEN"].lower().split()
+
+ def map_to_use_expand(myvals):
+ ret = {}
+ for exp in use_expand:
+ ret[exp] = []
+ for val in myvals[:]:
+ if val.startswith(exp.lower()+"_"):
+ ret[exp].append(val[len(exp)+1:])
+ myvals.remove(val)
+ ret["USE"] = myvals
+ for exp in use_expand_hidden:
+ if exp in ret:
+ del ret[exp]
+ return ret
+
+ cur_iuse_map = map_to_use_expand(cur_iuse)
+ cur_use_map = map_to_use_expand(cur_use)
+ old_iuse_map = map_to_use_expand(old_iuse)
+ old_use_map = map_to_use_expand(old_use)
+
+ use_expand.sort()
+ use_expand.insert(0, "USE")
+
+ for key in use_expand:
+ verboseadd += create_use_string(key.upper(), cur_iuse_map[key], cur_use_map[key],
+ old_iuse_map[key], cur_use_map[key], is_new)
+ if "--verbose" in myopts:
# size verbose
mysize=0
if x[0] == "ebuild" and x[-1]!="nomerge":
if ("--skipfirst" in myopts) and mymergelist:
del portage.mtimedb["resume"]["mergelist"][0]
del mymergelist[0]
+ for bigkey in mymergelist:
+ (pkgtype, root, cpv, action) = bigkey
+ if pkgtype == "binary" and not portage.db["/"]["bintree"].dbapi.match("="+cpv) or \
+ pkgtype == "ebuild" and not portage.db["/"]["porttree"].dbapi.xmatch("match-all", "="+cpv):
+ print red("!!! Error: The resume list contains packages that no longer")
+ print red("!!! available to be emerged. Please restart/continue")
+ print red("!!! the merge operation manually.")
+ sys.exit(1)
else:
myfavs=portage.grabfile(portage.root+portage.WORLD_FILE)
myfavdict=genericdict(myfavs)
self.pkgsettings["FEATURES"]=string.join(myfeat)
+ if "parallel-fetch" in myfeat and not ("--ask" in myopts or "--pretend" in myopts or "--fetchonly" in myopts):
+ if "distlocks" not in myfeat:
+ print red("!!!")
+ print red("!!!")+" parallel-fetching requires the distlocks feature enabled"
+ print red("!!!")+" you have it disabled, thus parallel-fetching is being disabled"
+ print red("!!!")
+ elif len(mymergelist) > 1:
+ print ">>> starting parallel fetching"
+ pid = os.fork()
+ if not pid:
+ sys.stdin.close()
+ sys.stdout.close()
+ sys.stderr.close()
+ sys.stdout = open("/dev/null","w")
+ sys.stderr = open("/dev/null","w")
+ os.dup2(sys.stdout.fileno(), 1)
+ os.dup2(sys.stdout.fileno(), 2)
+ for x in ("autoaddcvs", "cvs"):
+ try: myfeat.remove(x)
+ except ValueError: pass
+ self.pkgsettings["FEATURES"] = " ".join(myfeat)
+ ret = 0
+ for x in mymergelist:
+ if x[0] != "ebuild":
+ continue
+ try:
+ ret = portage.doebuild(portage.portdb.findname(x[2]), "fetch", x[1], self.pkgsettings,
+ cleanup=0, fetchonly=True, tree="porttree")
+ except SystemExit:
+ raise
+ except Exception:
+ ret = 1
+ sys.exit(0)
+ portage.portage_exec.spawned_pids.append(pid)
+
mergecount=0
for x in mymergelist:
mergecount+=1
if portage.db[portage.root]["bintree"].isremote(x[2]):
short_msg = "emerge: ("+str(mergecount)+" of "+str(len(mymergelist))+") "+x[pkgindex]+" Fetch"
emergelog(" --- ("+str(mergecount)+" of "+str(len(mymergelist))+") Fetching Binary ("+x[pkgindex]+"::"+mytbz2+")", short_msg=short_msg)
- portage.db[portage.root]["bintree"].gettbz2(x[2])
+ if not portage.db[portage.root]["bintree"].gettbz2(x[2]):
+ sys.exit(1)
if ("--fetchonly" in myopts) or ("--fetch-all-uri" in myopts):
continue
chk_updated_cfg_files()
- if(mybestpv != mypvs):
+ if(mybestpv != mypvs) and not "--quiet" in myopts:
print
print red(" * ")+bold("An update to portage is available.")+" It is _highly_ recommended"
print red(" * ")+"that you update portage now, before any other packages are updated."
if ("--resume" in myopts) and portage.mtimedb.has_key("resume"):
myresumeopts=portage.mtimedb["resume"]["myopts"][:]
- while "--skipfirst" in myresumeopts:
- myresumeopts.remove("--skipfirst")
- while "--ask" in myresumeopts:
- myresumeopts.remove("--ask")
+ for opt in ("--skipfirst", "--ask", "--verbose"):
+ while opt in myresumeopts:
+ myresumeopts.remove(opt)
for myopt in myopts:
if myopt not in myresumeopts:
if os.path.exists(self.destpath) and self.md5sum:
ok,reason = portage_checksum.verify_all(self.destpath, md5sum)
if not ok:
- portage_util.writemsg("Failed verification:" + reason + "\n")
+ portage_util.writemsg("Failed verification:"+reason[0]+" (got "+reason[1]+", expected "+reason[2]+"\n")
return 1
return 0
#!@BASH@
-# Copyright 1999-2004 Gentoo Foundation
+# Copyright 1999-2005 Gentoo Foundation
# Distributed under the terms of the GNU General Public License v2
-# $Id: /var/cvsroot/gentoo-src/portage/bin/prepall,v 1.10 2004/10/14 23:31:33 ferringb Exp $
+# $Id: prepall 2230 2005-11-01 01:35:59Z vapier $
prepallman
prepallinfo
prepallstrip
-#this should help to ensure that all (most?) shared libraries are executable
+# this should help to ensure that all (most?) shared libraries are executable
+# and that all libtool scripts / static libraries are not executable
for i in "${D}"opt/*/lib{,32,64} \
"${D}"lib{,32,64} \
"${D}"usr/lib{,32,64} \
"${D}"usr/X11R6/lib{,32,64} ; do
- [ ! -d "${i}" ] && continue
+ [[ ! -d ${i} ]] && continue
for j in "${i}"/*.so.* "${i}"/*.so ; do
- [ ! -e "${j}" ] && continue
- [ -L "${j}" ] && continue
+ [[ ! -e ${j} ]] && continue
+ [[ -L ${j} ]] && continue
+ [[ -x ${j} ]] && continue
echo "making executable: /${j/${D}/}"
chmod +x "${j}"
done
+
+ for j in "${i}"/*.a "${i}"/*.la ; do
+ [[ ! -e ${j} ]] && continue
+ [[ -L ${j} ]] && continue
+ [[ ! -x ${j} ]] && continue
+ echo "removing executable bit: /${j/${D}/}"
+ chmod -x "${j}"
+ done
+done
+
+# When installing static libraries into /usr/lib and shared libraries into
+# /lib, we have to make sure we have a linker script in /usr/lib along side
+# the static library, or gcc will utilize the static lib when linking :(.
+# http://bugs.gentoo.org/4411
+for a in "${D}"usr/lib*/*.a ; do
+ s=${a%.a}.so
+ if [[ ! -e ${s} ]] ; then
+ s=${s%usr/*}${s##*/usr/}
+ if [[ -e ${s} ]] ; then
+ echo -e "\aQA Notice: missing gen_usr_ldscript for ${s##*/}\a"
+ sleep 1
+ fi
+ fi
done
-# Move aclocals
-for i in `find "${D}"/ -name "aclocal" -type d 2>/dev/null` ; do
- [ -z "${i}" ] && continue
+# Make sure people don't store libtool files or static libs in /lib
+f=$(ls "${D}"lib*/*.{a,la} 2>/dev/null)
+if [[ -n ${f} ]] ; then
+ echo -e "\n\aQA Notice: excessive files found in the / partition\a"
+ echo "${f}"
+ sleep 1
+fi
- # Strip double '/'
- dir1="`echo "${i}" | sed -e 's:/\{2,\}:/:g'`"
- dir2="`echo "${D}/usr/share/aclocal" | sed -e 's:/\{2,\}:/:g'`"
-
- [ "${dir1}" == "${dir2}" ] && continue
+# Verify that the libtool files don't contain bogus $D entries.
+for a in "${D}"usr/lib*/*.la ; do
+ s=${a##*/}
+ if grep -qs "${D}" "${a}" ; then
+ echo -e "\n\aQA Notice: ${s} appears to contain PORTAGE_TMPDIR paths\a"
+ sleep 1
+ fi
+done
+
+if type -p scanelf > /dev/null ; then
- echo "moving aclocal: /${i/${D}/}"
- install -d "${D}"usr/share/aclocal
- mv "${i}"/* "${D}"usr/share/aclocal
- rm -fr "${i}"
+# Run some sanity checks on shared libraries
+for d in "${D}"lib* "${D}"usr/lib* ; do
+ f=$(scanelf -ByF '%S %p' "${d}"/lib*.so* | gawk '$2 == "" { print }')
+ if [[ -n ${f} ]] ; then
+ echo -e "\n\aQA Notice: the following shared libraries lack a SONAME\a"
+ echo "${f}"
+ sleep 1
+ fi
+
+ f=$(scanelf -ByF '%n %p' "${d}"/lib*.so* | gawk '$2 == "" { print }')
+ if [[ -n ${f} ]] ; then
+ echo -e "\n\aQA Notice: the following shared libraries lack NEEDED entries\a"
+ echo "${f}"
+ sleep 1
+ fi
done
+
+fi
# Distributed under the terms of the GNU General Public License v2
# $Id: /var/cvsroot/gentoo-src/portage/bin/preplib,v 1.7.2.1 2005/01/02 09:36:02 jstubbs Exp $
+ewarn "QA Notice: Please stop using me, fix your package instead"
+
LIBDIR_VAR="LIBDIR_${ABI}"
if [ -n "${ABI}" -a -n "${!LIBDIR_VAR}" ]; then
CONF_LIBDIR="${!LIBDIR_VAR}"
+++ /dev/null
-#!@BASH@
-# Copyright 1999-2004 Gentoo Foundation
-# Distributed under the terms of the GNU General Public License v2
-# $Id: /var/cvsroot/gentoo-src/portage/bin/preplib.so,v 1.10.2.1 2004/12/06 03:01:43 carpaski Exp $
-
-for x in "$@" ; do
- if [ -d "${D}${x}" ] ; then
- ldconfig -n -N "${D}${x}"
- fi
-done
if [ "${FEATURES//*nostrip*/true}" == "true" ] || [ "${RESTRICT//*nostrip*/true}" == "true" ] ; then
echo "nostrip"
- STRIP="/bin/false"
- PORTAGE_STRIP_FLAGS=""
-else
- STRIP=${STRIP:-${CHOST}-strip}
- type -p -- ${STRIP} > /dev/null || STRIP=strip
- PORTAGE_STRIP_FLAGS=${PORTAGE_STRIP_FLAGS:---strip-unneeded}
+ exit 0
fi
+STRIP="${STRIP:-${CHOST}-strip}"
+type -p -- ${STRIP} > /dev/null || STRIP=strip
+
+PORTAGE_STRIP_FLAGS=${PORTAGE_STRIP_FLAGS:---strip-unneeded}
+
banner=1
-retval=0
-
-for x in "$@" ; do
- if [ -d "${x}" ]; then
- # We only want files. So make a pass for each directory and call again.
- find "${x}" -type f \( -perm -0100 -or -perm -0010 -or -perm -0001 -or -name '*.so' -or -name '*.so.*' \) -print0 |
- $XARGS -0 -n500 prepstrip
- else
- if [ ${banner} -eq 1 ] ; then
- echo "strip: ${STRIP} ${PORTAGE_STRIP_FLAGS}"
- banner=0
- fi
-
- f=$(file "${x}") || continue
- [ -z "${f}" ] && continue
-
- if [ -z "${f/*current ar archive*/}" ]; then
- echo " ${x:${#D}:${#x}}"
- ${STRIP} -g "${x}"
- fi
- if [ -z "${f/*SB executable*/}" ]; then
- echo " ${x:${#D}:${#x}}"
- ${STRIP} "${x}"
- fi
- if [ -z "${f/*SB shared object*/}" ]; then
- echo " ${x:${#D}:${#x}}"
- ${STRIP} ${PORTAGE_STRIP_FLAGS} "${x}"
- fi
+
+save_elf_debug() {
+ local x=$1
+ local y="${D}/usr/lib/debug/${x:${#D}:${#x}}"
+
+ [ "${FEATURES//*splitdebug*/true}" != "true" ] && return 0
+
+ mkdir -p $(dirname ${y})
+ ${CHOST}-objcopy --only-keep-debug ${x} ${y}.debug
+ ${CHOST}-objcopy --add-gnu-debuglink=${y}.debug ${x}
+}
+
+for x in $(scanelf -yRBF%F "$@") $(for y in "$@"; do find $y -type f -name '*.a' -print0 ; done); do
+ if [ ${banner} -eq 1 ] ; then
+ echo "strip: ${STRIP} ${PORTAGE_STRIP_FLAGS}"
+ banner=0
fi
-done
-exit ${retval}
+ f=$(file "${x}") || continue
+ [ -z "${f}" ] && continue
+
+ if [[ ${f} == *"current ar archive"* ]] ; then
+ echo " ${x:${#D}:${#x}}"
+ ${STRIP} -g "${x}"
+ fi
+ if [[ ${f} == *"SB executable"* || ${f} == *"SB shared object"* ]] ; then
+ echo " ${x:${#D}:${#x}}"
+ save_elf_debug "${x}"
+ ${STRIP} ${PORTAGE_STRIP_FLAGS} "${x}"
+ fi
+done
# after a merge -- for debugging purposes only.
# 'nostrip' prevents the stripping of binaries.
# 'notitles' disables xterm titlebar updates (which contain status info).
+# 'parallel-fetch'
+# do fetching in parallel to compilation
# 'sandbox' enables sandboxing when running emerge and ebuild.
# 'strict' causes portage to react strongly to conditions that are
# potentially dangerous, like missing/incorrect Manifest files.
# ${PACKAGE} - expands to the cpv entry of the processed
# package (see $PVR in ebuild(5))
# ${LOGFILE} - absolute path to the logfile
-#PORTAGE_ELOG_COMMAND="/path/to/logprocessor -p ${PACKAGE} -f ${LOGFILE}"
+# Both variables have to be quoted with single quotes
+#PORTAGE_ELOG_COMMAND="/path/to/logprocessor -p '${PACKAGE}' -f '${LOGFILE}'"
# PORTAGE_ELOG_MAILURI: this variable holds all important settings for the mail
# module. In most cases listing the recipient adress and
# ${PACKAGE} - expands to the cpv entry of the processed
# package (see $PVR in ebuild(5))
# ${LOGFILE} - absolute path to the logfile
-#PORTAGE_ELOG_COMMAND="/path/to/logprocessor -p ${PACKAGE} -f ${LOGFILE}"
+# Both variables have to be quoted with single quotes
+#PORTAGE_ELOG_COMMAND="/path/to/logprocessor -p '${PACKAGE}' -f '${LOGFILE}'"
# PORTAGE_ELOG_MAILURI: this variable holds all important settings for the mail
# module. In most cases listing the recipient adress and
# ${PACKAGE} - expands to the cpv entry of the processed
# package (see $PVR in ebuild(5))
# ${LOGFILE} - absolute path to the logfile
-#PORTAGE_ELOG_COMMAND="/path/to/logprocessor -p ${PACKAGE} -f ${LOGFILE}"
+# Both variables have to be quoted with single quotes
+#PORTAGE_ELOG_COMMAND="/path/to/logprocessor -p '${PACKAGE}' -f '${LOGFILE}'"
# PORTAGE_ELOG_MAILURI: this variable holds all important settings for the mail
# module. In most cases listing the recipient adress and
# ${PACKAGE} - expands to the cpv entry of the processed
# package (see $PVR in ebuild(5))
# ${LOGFILE} - absolute path to the logfile
-#PORTAGE_ELOG_COMMAND="/path/to/logprocessor -p ${PACKAGE} -f ${LOGFILE}"
+# Both variables have to be quoted with single quotes
+#PORTAGE_ELOG_COMMAND="/path/to/logprocessor -p '${PACKAGE}' -f '${LOGFILE}'"
# PORTAGE_ELOG_MAILURI: this variable holds all important settings for the mail
# module. In most cases listing the recipient adress and
# ${PACKAGE} - expands to the cpv entry of the processed
# package (see $PVR in ebuild(5))
# ${LOGFILE} - absolute path to the logfile
-#PORTAGE_ELOG_COMMAND="/path/to/logprocessor -p ${PACKAGE} -f ${LOGFILE}"
+# Both variables have to be quoted with single quotes
+#PORTAGE_ELOG_COMMAND="/path/to/logprocessor -p '${PACKAGE}' -f '${LOGFILE}'"
# PORTAGE_ELOG_MAILURI: this variable holds all important settings for the mail
# module. In most cases listing the recipient adress and
# ${PACKAGE} - expands to the cpv entry of the processed
# package (see $PVR in ebuild(5))
# ${LOGFILE} - absolute path to the logfile
-#PORTAGE_ELOG_COMMAND="/path/to/logprocessor -p ${PACKAGE} -f ${LOGFILE}"
+# Both variables have to be quoted with single quotes
+#PORTAGE_ELOG_COMMAND="/path/to/logprocessor -p '${PACKAGE}' -f '${LOGFILE}'"
# PORTAGE_ELOG_MAILURI: this variable holds all important settings for the mail
# module. In most cases listing the recipient adress and
# ${PACKAGE} - expands to the cpv entry of the processed
# package (see $PVR in ebuild(5))
# ${LOGFILE} - absolute path to the logfile
-#PORTAGE_ELOG_COMMAND="/path/to/logprocessor -p ${PACKAGE} -f ${LOGFILE}"
+# Both variables have to be quoted with single quotes
+#PORTAGE_ELOG_COMMAND="/path/to/logprocessor -p '${PACKAGE}' -f '${LOGFILE}'"
# PORTAGE_ELOG_MAILURI: this variable holds all important settings for the mail
# module. In most cases listing the recipient adress and
# ${PACKAGE} - expands to the cpv entry of the processed
# package (see $PVR in ebuild(5))
# ${LOGFILE} - absolute path to the logfile
-#PORTAGE_ELOG_COMMAND="/path/to/logprocessor -p ${PACKAGE} -f ${LOGFILE}"
+# Both variables have to be quoted with single quotes
+#PORTAGE_ELOG_COMMAND="/path/to/logprocessor -p '${PACKAGE}' -f '${LOGFILE}'"
# PORTAGE_ELOG_MAILURI: this variable holds all important settings for the mail
# module. In most cases listing the recipient adress and
# ${PACKAGE} - expands to the cpv entry of the processed
# package (see $PVR in ebuild(5))
# ${LOGFILE} - absolute path to the logfile
-#PORTAGE_ELOG_COMMAND="/path/to/logprocessor -p ${PACKAGE} -f ${LOGFILE}"
+# Both variables have to be quoted with single quotes
+#PORTAGE_ELOG_COMMAND="/path/to/logprocessor -p '${PACKAGE}' -f '${LOGFILE}'"
# PORTAGE_ELOG_MAILURI: this variable holds all important settings for the mail
# module. In most cases listing the recipient adress and
# ${PACKAGE} - expands to the cpv entry of the processed
# package (see $PVR in ebuild(5))
# ${LOGFILE} - absolute path to the logfile
-#PORTAGE_ELOG_COMMAND="/path/to/logprocessor -p ${PACKAGE} -f ${LOGFILE}"
+# Both variables have to be quoted with single quotes
+#PORTAGE_ELOG_COMMAND="/path/to/logprocessor -p '${PACKAGE}' -f '${LOGFILE}'"
# PORTAGE_ELOG_MAILURI: this variable holds all important settings for the mail
# module. In most cases listing the recipient adress and
# ${PACKAGE} - expands to the cpv entry of the processed
# package (see $PVR in ebuild(5))
# ${LOGFILE} - absolute path to the logfile
-#PORTAGE_ELOG_COMMAND="/path/to/logprocessor -p ${PACKAGE} -f ${LOGFILE}"
+# Both variables have to be quoted with single quotes
+#PORTAGE_ELOG_COMMAND="/path/to/logprocessor -p '${PACKAGE}' -f '${LOGFILE}'"
# PORTAGE_ELOG_MAILURI: this variable holds all important settings for the mail
# module. In most cases listing the recipient adress and
# ${PACKAGE} - expands to the cpv entry of the processed
# package (see $PVR in ebuild(5))
# ${LOGFILE} - absolute path to the logfile
-#PORTAGE_ELOG_COMMAND="/path/to/logprocessor -p ${PACKAGE} -f ${LOGFILE}"
+# Both variables have to be quoted with single quotes
+#PORTAGE_ELOG_COMMAND="/path/to/logprocessor -p '${PACKAGE}' -f '${LOGFILE}'"
# PORTAGE_ELOG_MAILURI: this variable holds all important settings for the mail
# module. In most cases listing the recipient adress and
# ${PACKAGE} - expands to the cpv entry of the processed
# package (see $PVR in ebuild(5))
# ${LOGFILE} - absolute path to the logfile
-#PORTAGE_ELOG_COMMAND="/path/to/logprocessor -p ${PACKAGE} -f ${LOGFILE}"
+# Both variables have to be quoted with single quotes
+#PORTAGE_ELOG_COMMAND="/path/to/logprocessor -p '${PACKAGE}' -f '${LOGFILE}'"
# PORTAGE_ELOG_MAILURI: this variable holds all important settings for the mail
# module. In most cases listing the recipient adress and
AC_HEADER_DIRENT
AC_HEADER_STDC
AC_HEADER_SYS_WAIT
-AC_CHECK_HEADERS(fcntl.h limits.h strings.h sys/file.h sys/time.h unistd.h getopt.h)
+AC_CHECK_HEADERS(fcntl.h limits.h strings.h sys/file.h sys/time.h unistd.h getopt.h stdio.h)
dnl Checks for typedefs, structures, and compiler characteristics.
AC_C_CONST
esac],
[wheelgroup="wheel"])
+AC_ARG_WITH(wheelgid,
+AC_HELP_STRING([--with-wheelgid=wheelgid],[let portage use 'wheelgid' as wheel group (default 0)]),
+[case "${withval}" in
+ ""|yes) AC_MSG_ERROR(bad value ${withval} for --with-wheelgid);;
+ *) wheelgid="${withval}";;
+esac],
+[wheelgid="0"])
+
AC_ARG_WITH(rootuser,
AC_HELP_STRING([--with-rootuser=me],[allows 'me' to do root tasks (default root)]),
[case "${withval}" in
-.TH "DISPATCH-CONF" "1" "May 2004" "Portage 2.0.51" "Portage"
+.TH "DISPATCH-CONF" "1" "Dec 2005" "Portage 2.1" "Portage"
.SH NAME
dispatch-conf \- Sanely update configuration files after emerging new packages
.SH SYNOPSIS
-.TH "EBUILD" "1" "Nov 2002" "Portage 2.0.51" "Portage"
+.TH "EBUILD" "1" "Dec 2005" "Portage 2.1" "Portage"
.SH "NAME"
ebuild \- manual page for the ebuild program, a low level interface to
the Portage system.
-.TH "EBUILD" "5" "Feb 2003" "Portage 2.0.51" "portage"
+.TH "EBUILD" "5" "Dec 2005" "Portage 2.1" "portage"
.SH "NAME"
ebuild \- the internal format, variables, and functions in an ebuild script
.SH "DESCRIPTION"
.br
You may use the same syntax to vary dependencies as seen above in \fBDEPEND\fR.
.TP
-\fBRESTRICT\fR = \fI[fetch,mirror,nostrip,sandbox,test,userpriv]\fR
+\fBRESTRICT\fR = \fI[strip,mirror,fetch,userpriv]\fR
This should be a space delimited list of portage features to restrict.
.PD 0
.RS
.I primaryuri
fetch from URL's in \fBSRC_URI\fR before \fBGENTOO_MIRRORS\fR.
.TP
-.I nostrip
+.I strip
final binaries/libraries will not be stripped of debug symbols.
.TP
.I sandbox
This function can be used if the package needs specific setup actions or
checks to be preformed before anything else.
.br
-Initial working directory of $PORTAGE_TMPDIR.
+Initial working directory of ${PORTAGE_TMPDIR}.
.TP
.B src_unpack
This function is used to unpack all the sources in \fIA\fR to \fIWORKDIR\fR.
.fi
.RE
.TP
-\fBuse_with\fR \fI<USE item>\fR \fI[configure option]\fR
-Useful for creating custom options to pass to a configure script. If
-\fIUSE item\fR is in the \fBUSE\fR variable, then the string
-\fI--with-[configure option]\fR will be echoed. If \fIUSE item\fR is
-not in the \fBUSE\fR variable, then the string
-\fI--without-[configure option]\fR will be echoed. If
-\fIconfigure option\fR is not specified, then \fIUSE item\fR will be
-used in its place.
+\fBuse_with\fR \fI<USE item>\fR \fI[configure name]\fR \fI[configure opt]\fR
+Useful for creating custom options to pass to a configure script. If \fIUSE
+item\fR is in the \fBUSE\fR variable and a \fIconfigure opt\fR is specified,
+then the string \fI--with-[configure name]=[configure opt]\fR will be echoed.
+If \fIconfigure opt\fR is not specified, then just \fI--with-[configure
+name]\fR will be echoed. If \fIUSE item\fR is not in the \fBUSE\fR variable,
+then the string \fI--without-[configure name]\fR will be echoed. If
+\fIconfigure name\fR is not specified, then \fIUSE item\fR will be used in
+its place.
.RS
.TP
-.I Example:
+.I Examples:
.nf
+USE="opengl"
+myconf=$(use_with opengl)
+(myconf now has the value "--with-opengl")
+
USE="jpeg"
-myconf="$(use_with jpeg libjpeg)"
+myconf=$(use_with jpeg libjpeg)
(myconf now has the value "--with-libjpeg")
USE=""
-myconf="$(use_with jpeg libjpeg)"
+myconf=$(use_with jpeg libjpeg)
(myconf now has the value "--without-libjpeg")
-USE="opengl"
-myconf="$(use_with opengl")
-(myconf now has the value "--with-opengl")
+USE="sdl"
+myconf=$(use_with sdl SDL all-plugins)
+(myconf now has the value "--with-SDL=all-plugins")
.fi
.RE
.TP
-\fBuse_enable\fR \fI<USE item>\fR \fI[configure option]\fR
-Useful for creating custom options to pass to a configure script. If
-\fIUSE item\fR is in the \fBUSE\fR variable, then the string
-\fI--enable-[configure option]\fR will be echoed. If \fIUSE item\fR is
-not in the \fBUSE\fR variable, then the string
-\fI--disable-[configure option]\fR will be echoed. If \fIconfigure option\fR
-is not specified, then \fIUSE item\fR will be used in its place.
-.br
-See \fBuse_with\fR for an example.
+\fBuse_enable\fR \fI<USE item>\fR \fI[configure name]\fR \fI[configure opt]\fR
+Same as \fBuse_with\fR above, except that the configure options are
+\fI--enable-\fR instead of \fI--with-\fR and \fI--disable-\fR instead of
+\fI--without-\fR.
.TP
\fBhas\fR \fI<item>\fR \fI<item list>\fR
If \fIitem\fR is in \fIitem list\fR, then \fIitem\fR is echoed and \fBhas\fR
.TP
\fBprepinfo\fR \fI[dir]\fR
.TP
-\fBpreplib\fR \fI[dir]\fR
-.TP
-\fBpreplib.so\fR \fI[dir]\fR
-.TP
\fBprepman\fR \fI[dir]\fR
.TP
\fBprepstrip\fR \fI[dir]\fR
\fIusr\fR. \fBprepinfo\fR will then compress all the files in
${D}/\fIdir\fR/info.
.TP
-.B preplib:
-If a \fIdir\fR is not specified, then \fBpreplib\fR will assume the dir
-\fIusr\fR. \fBpreplib\fR will then run 'ldconfig -n -N' on ${D}/\fIdir\fR/lib.
-.TP
-.B preplib.so:
-All the files with '.so' in their name and are found in ${D}/\fIdir\fR will
-be stripped of their debug symbols. You may specify multiple directories.
-.TP
.B prepman:
If a \fIdir\fR is not specified, then \fBprepman\fR will assume the dir
\fIusr\fR. \fBprepman\fR will then compress all the files in
.TP
\fBdoman\fR \fI[\-i18n=<locale>]\fR \fI<man-page> [list of more man\-pages]\fR
Installs manual\-pages into /usr/share/man/man[0\-9n] depending on the
-manual file ending. The files are gzipped if they are not already. You can
-specify locale-specific manpages with the \fI\-i18n\fR option. Then the
-man-page will be installed into /usr/share/man/\fI<locale>\fR/man[0\-9n].
-Creates all necessary dirs.
-
+manual file ending. The files are compressed if they are not already. You
+can specify locale-specific manpages with the \fI\-i18n\fR option. Then the
+man-page will be installed into /usr/share/man/\fI<locale>\fR/man[0\-9n].
.PD 0
.TP
\fBdohard\fR \fI<filename> <linkname>\fR
.TP
\fBdohtml\fR \fI [\-a filetypes] [\-r] [\-x list\-of\-dirs\-to\-ignore] [list\-of\-files\-and\-dirs]\fR
Installs the files in the list of files (space\-separated list) into
-/usr/share/doc/${PF}/html provided the file ends in .html, .png, .js, .jpg,
+/usr/share/doc/${PF}/html provided the file ends in .html, .png, .js, .jpg,
or .css. Setting \fI\-a\fR limits what types of files will be included,
-\fI\-A\fR appends to the default list, setting \fI\-x\fR sets which dirs to
+\fI\-A\fR appends to the default list, setting \fI\-x\fR sets which dirs to
exclude (CVS excluded by default), \fI\-r\fR sets recursive.
.TP
\fBdoinfo\fR \fI<info-file> [list of more info\-files]\fR
-.TH "EMAINT" "1" "Nov 2005" "Portage 2.0.53" "Portage"
+.TH "EMAINT" "1" "Dec 2005" "Portage 2.1" "Portage"
.SH NAME
emaint \- updates environment settings automatically
.SH SYNOPSIS
-.TH "EMERGE" "1" "Jun 2003" "Portage 2.0.51" "Portage"
+.TH "EMERGE" "1" "Dec 2005" "Portage 2.1" "Portage"
.SH "NAME"
emerge \- Command\-line interface to the Portage system
.SH "SYNOPSIS"
information when submitting a bug report.\fR Expanded output can be obtained
with the \fI\-\-verbose\fR option.
.TP
+.BR "\-\-inject " (\fB\-i\fR)
+Injecting a package inserts a 'stub' for that package so that Portage
+thinks that it is installed. It is handy if you need, say, a binary
+version of XFree86 for esoteric hardware, or you just like to roll
+your own packages. You \fBmust\fR specify a category and particular
+version of a package for injecting. For example,
+\fBemerge \-\-inject sys\-kernel/gentoo\-sources\-2.4.19\fR.
+.TP
.BR \-\-metadata
Causes portage to process all the metacache files as is normally done on the
tail end of an rsync update using \fBemerge \-\-sync\fR. The processing
.SH "OPTIONS"
.TP
.BR "\-\-ask " (\fB\-a\fR)
-Before performing the merge, display what ebuilds and tbz2s will be
-installed, in the same format as when using \fB\-\-pretend\fR; then ask
-whether to continue with the merge or abort. Using \fB\-\-ask\fR is more
-efficient than using \fB\-\-pretend\fR and then executing the same command
-without \fB\-\-pretend\fR, as dependencies will only need to be calculated
-once.
+Before performing the merge, display what ebuilds and tbz2s will be installed,
+in the same format as when using \fB\-\-pretend\fR; then ask whether to
+continue with the merge or abort. Using \fB\-\-ask\fR is more efficient than
+using \fB\-\-pretend\fR and then executing the same command without
+\fB\-\-pretend\fR, as dependencies will only need to be calculated once.
.TP
.BR "\-\-buildpkg " (\fB\-b\fR)
Tells emerge to build binary packages for all ebuilds processed in
listed in the dependencies of a package.
.TP
.BR "\-\-emptytree " (\fB\-e\fR)
-Virtually tweaks the tree of installed packages to contain nothing. This is
-great to use together with \fI\-\-pretend\fR. This makes it possible for
-developers to get a complete overview of the entire dependency tree of a
-certain package.
+Reinstalls all world packages and their dependencies to the current USE
+specifications while differing from the installed set of packages as
+little as possible. You should run with \fB\-\-pretend\fR first to make
+sure the result is what you expect.
.TP
.BR "\-\-fetchonly " (\fB\-f\fR)
Instead of doing any package building, just perform fetches for all
compilation. An asterisk marks when a USE flag has changed since the package
was compiled.
.TP
-.BR \-\-nocolor
+.BR "\-\-nocolor "
Make sure none of the output from portage contains color.
.TP
.BR "\-\-noconfmem"
.br
\fIN\fR = new, (not yet installed)
.br
-\fIS\fR = new, slot installation (side-by-side versions)
+\fIS\fR = new, SLOT installation (side-by-side versions)
.br
\fIU\fR = updating, (changing versions)
.br
use this option only in combination with the world or system target.
.TP
.BR "\-\-upgradeonly " (\fB\-U\fR)
-Updates packages, but excludes updates that would result in a lower version
-of the package being installed. \fBSLOT\fRs are considered at a basic level.
+Updates packages, but excludes updates that would result in a lower version of
+the package being installed. \fBSLOT\fRs are considered at a basic level.
.br
This option is deprecated and should not be used anymore. Please use the
/etc/portage/package.* files from now on.
.BR "\-\-version " (\fB\-V\fR)
Displays the version number of \fBemerge\fR. It cannot be used in
conjunction with other options.
+.SH "ENVIRONMENT OPTIONS"
+.TP
+\fBROOT\fR = \fI[path]\fR
+Use \fBROOT\fR to specify the target root filesystem to be used for
+merging packages or ebuilds.
+.br
+Defaults to /.
.SH "OUTPUT"
When utilizing \fBemerge\fR with the \fB\-\-pretend\fR and \fB\-\-verbose\fR
flags, the output may be a little hard to understand at first. This section
-.TH "ENV-UPDATE" "1" "Feb 2003" "Portage 2.0.51" "Portage"
+.TH "ENV-UPDATE" "1" "Dec 2005" "Portage 2.1" "Portage"
.SH NAME
env-update \- updates environment settings automatically
.SH SYNOPSIS
-.TH "ETC-UPDATE" "1" "Feb 2003" "Portage 2.0.51" "Portage"
+.TH "ETC-UPDATE" "1" "Dec 2005" "Portage 2.1" "Portage"
.SH NAME
etc-update \- handle configuration file updates
.SH SYNOPSIS
-.TH "MAKE.CONF" "5" "Feb 2003" "Portage 2.0.51" "portage"
+.TH "MAKE.CONF" "5" "Dec 2005" "Portage 2.1" "portage"
.SH "NAME"
make.conf \- custom settings for Portage
.SH "SYNOPSIS"
from /etc/make.profile/make.defaults. Please note that all user
settings should be made in the environment or in /etc/make.conf,
which is intended to be customized by the user.
-.br
+.br
Exceptions are incremental variables such as USE, CONFIG_PROTECT*,
and ACCEPT_KEYWORDS. Incremental variables are propagated down from
make.defaults to make.globals to make.conf to the environment
of the 'x86' architecture would set this to '~x86' while ppc users would
set this to '~ppc'. This is an incremental variable. Only define a
~arch.
-.br
+.br
Defaults to the value of $ARCH.
.TP
-\fBAUTOCLEAN\fR = \fI["yes" | "no"]\fR
-Automatically cleans the system by removing outdated packages which will not
-remove functionalities or prevent your system from working. On major ABI
-changes this may need to be set to off to ensure that the system can be
-rebuilt using the new libs before the old ones are removed. Downgrading with
-this option turned off may result in missing symlinks and an inoperable system.
-.br
-Defaults to yes.
-.TP
\fBBUILD_PREFIX\fR = \fI[path]\fR
Defines the location of the package working directory.
-.br
+.br
Defaults to ${PORTAGE_TMPDIR}/portage
.TP
.B CBUILD
This variable is passed by the \fIebuild scripts\fR to the \fIconfigure\fR
-as \fI\-\-build=${CBUILD}\fR only if it is defined.
+as \fI\-\-build=${CBUILD}\fR only if it is defined. Do not set this yourself
+unless you know what you are doing.
.TP
\fBCCACHE_SIZE\fR = \fI"size"\fR
This controls the space use limitations for ccache. The default is 2 gigabytes
to your specifications. Please use sane settings as some packages will fail to
compile/run if the optimizations are too extreme.
-For more information:
-.br
-http://gcc.gnu.org/onlinedocs/gcc\-3.2/gcc/Invoking\-GCC.html
-.br
-http://gcc.gnu.org/onlinedocs/gcc\-3.3/gcc/Invoking\-GCC.html
-.br
-http://gcc.gnu.org/onlinedocs/gcc\-2.95.3/gcc_2.html
+For more information, see the \fIInvoking GCC\fR section of the gcc manual:
+.br
+http://gcc.gnu.org/onlinedocs/
.TP
.B CHOST
This variable is passed by the \fIebuild scripts\fR to the \fIconfigure\fR
step as \fI\-\-host=${CHOST}\fR. This way you can force the build\-host.
For more information:
-.br
+.br
http://gcc.gnu.org/onlinedocs/gcc\-3.2/gcc/Submodel\-Options.html
-.br
+.br
http://gcc.gnu.org/onlinedocs/gcc\-3.3/gcc/Submodel\-Options.html
-.br
+.br
http://gcc.gnu.org/onlinedocs/gcc\-2.95.3/gcc_2.html
.TP
\fBCLEAN_DELAY\fR = \fIinteger\fR
Determines how long the countdown delay will be after running `emerge clean`.
-.br
+.br
Defaults to 5 seconds.
.TP
\fBCONFIG_PROTECT\fR = \fI[space delimited list of dirs]\fR
.TP
\fBDISTDIR\fR = \fI[path]\fR
Defines the location of your local source file repository.
-.br
+.br
Defaults to ${PORTDIR}/distfiles.
.TP
.B EBEEP_IGNORE
Portage uses lockfiles to ensure competing instances don't clobber
each other's files. This feature is enabled by default but may cause
heartache on less intelligent remote filesystems like NFSv2 and some
-strangely configured Samba server (oplocks off, NFS re-export). A tool
+strangely configured Samba server (oplocks off, NFS re\-export). A tool
@PORTAGE_BASE@bin/clean_locks exists to help handle lock issues
when a problem arises (normally due to a crash or disconnect).
.TP
.B notitles
Disables xterm titlebar updates (which contains status info).
.TP
+.B parallel-fetch
+Fetch in the background while compiling.
+.TP
.B sandbox
Enable sandbox\-ing when running \fBemerge\fR(1) and \fBebuild(1)\fR.
.TP
Enable the sandbox in the compile phase, when running without root privs (\fIuserpriv\fR).
.RE
.TP
-\fBFETCHCOMMAND\fR
-This variable contains the command used for fetching package\-sources from
-the internet.
+.B FETCHCOMMAND
+This variable contains the command used for fetching package sources from
+the internet. It must contain the full path to the executable as well as the
+place-holders \\${DISTDIR} and \\${URI}.
.TP
\fBGENTOO_MIRRORS\fR = \fI[URLs]\fR
Insert your space\-seperated list of local mirrors here. These
internet by \fBwget\fR(1). They are only required if you use a
proxy server for internet access.
.TP
-\fBMAKEOPTS\fR
+.B MAKEOPTS
Use this variable if you want to use parallel make. For example, if you
have a dual\-processor system, set this variable to "\-j2" or "\-j3" for
enhanced build performance with many packages. Suggested settings are
.TP
\fBNOCOLOR\fR = \fI["true" | "false"]\fR
Defines if color should be disabled by default.
-.br
+.br
Defaults to false.
.TP
\fBPKGDIR\fR = \fI[path]\fR
Defines the location where created .tbz2 binary packages will be stored.
-.br
+.br
Defaults to ${PORTDIR}/packages.
.TP
.B PORT_LOGDIR
.TP
\fBPORTAGE_TMPDIR\fR = \fI[path]\fR
Defines the location of the temporary build directories.
-.br
+.br
Defaults to /var/tmp.
.TP
\fBPORTDIR\fR = \fI[path]\fR
Defines the location of your Portage tree.
-.br
+.br
Defaults to /usr/portage.
.TP
\fBPORTDIR_OVERLAY\fR = \fI"[path] [different\-path] [etc...]"\fR
Defines the directories in which user made ebuilds may be stored and not
overwriten when `emerge \-\-sync` is run. This is a space delimited list of
directories.
-.br
+.br
Defaults to no value.
.TP
-\fBRESUMECOMMAND\fR
-This variable contains the command used for resuming package\-sources that
-have been partially downloaded by the \fBFETCHCOMMAND\fR.
+.B RESUMECOMMAND
+This variable contains the command used for resuming package sources that
+have been partially downloaded. It should be defined using the same format
+as \fBFETCHCOMMAND\fR.
.TP
\fBROOT\fR = \fI[path]\fR
Use \fBROOT\fR to specify the target root filesystem to be used for merging
Typically, you should set this setting in the environment rather than in
\fI/etc/make.conf\fR itself. It's commonly used for creating new build
images.
-.br
+.br
Defaults to /.
.TP
\fBRSYNC_EXCLUDEFROM\fR = \fI"/etc/portage/rsync_excludes"\fR
This is a file that portage will pass to rsync when it updates the portage
tree. Specific chucks of the tree may be excluded from the sync process.
This may cause dependency failures if you are not careful. The file format
-is pattern per line, blanks and ';' or '#' lines are comments. See
+is one pattern per line, blanks and ';' or '#' lines are comments. See
\fBrsync\fR(1) for more details.
.TP
\fBRSYNC_RETRIES\fR = \fI[NUMBER]\fR
The number of times rsync should retry on failed connections before
giving up.
-.br
+.br
Defaults to 3.
.TP
\fBRSYNC_TIMEOUT\fR = \fI[SECONDS]\fR
The number of seconds rsync should remain idle before it determines the
connection has timed out. Dialup users may need to set this value at or
above 300 seconds.
-.br
+.br
Defaults to 180 seconds.
.TP
\fBRPMDIR\fR = \fI[path]\fR
Defines the location where created RPM packages will be stored.
-.br
+.br
Defaults to ${PORTDIR}/rpm.
.TP
\fBSYNC\fR = \fI[RSYNC]\fR
Insert your preferred rsync mirror here. This rsync server
is used to sync the local portage tree when `emerge \-\-sync` is run.
-.br
+.br
Defaults to rsync://rsync.gentoo.org/gentoo\-portage
.TP
\fBUSE\fR = \fI[space delimited list of USE items]\fR
The above setting will cause the environment (env) to override per package (pkg)
settings to override make.conf (conf) to override auto generated values
from merged packages (auto) to override make.defaults (defaults).
-.br
+.br
\fB***warning***\fR
-.br
+.br
Do not modify this value unless you're a developer and you know what
you're doing. If you change this and something breaks, we will not help
you fix it.
-.br
+.br
Defaults to "env:pkg:conf:auto:defaults".
.SH "REPORTING BUGS"
-.TH "PORTAGE" "5" "Jan 2004" "Portage 2.0.51" "Portage"
+.TH "PORTAGE" "5" "Dec 2005" "Portage 2.1" "Portage"
.SH NAME
portage \- the heart of Gentoo
.SH "DESCRIPTION"
.I Format:
\- comments begin with #
\- one DEPEND atom per line followed by additional KEYWORDS
-\- blank lines imply unstable host arch
+\- lines without any KEYWORDS imply unstable host arch
.I Example:
# always use unstable libgd
app\-hackers
media\-other
.fi
+.RE
.TP
.BR /usr/portage/profiles/
Global Gentoo settings that are controlled by the developers. To override
-.TH "QUICKPKG" "1" "Jul 2003" "Portage 2.0.51" "Portage"
+.TH "QUICKPKG" "1" "Dec 2005" "Portage 2.1" "Portage"
.SH NAME
quickpkg \- creates portage packages
.SH SYNOPSIS
-.TH "REPOMAN" "1" "Nov 2002" "Portage 2.0.51" "Portage"
+.TH "REPOMAN" "1" "Dec 2005" "Portage 2.1" "Portage"
.SH NAME
repoman \- Gentoo's program to enforce a minimal level of quality assurance in packages added to the portage tree
.SH SYNOPSIS
autocommits = True
cleanse_keys = True
+ serialize_eclasses = False
def __init__(self, *args, **config):
super(database,self).__init__(*args, **config)
def iteritems(self):
return self.__db.iteritems()
- def __getitem__(self, cpv):
+ def _getitem(self, cpv):
# we override getitem because it's just a cpickling of the data handed in.
return pickle.loads(self.__db[cpv])
-
def _setitem(self, cpv, values):
self.__db[cpv] = pickle.dumps(values,pickle.HIGHEST_PROTOCOL)
def _delitem(self, cpv):
del self.__db[cpv]
-
def iterkeys(self):
return iter(self.__db)
-
def has_key(self, cpv):
return cpv in self.__db
-
def __del__(self):
if "__db" in self.__dict__ and self.__db != None:
self.__db.sync()
complete_eclass_entries_ = True
autocommits = False
cleanse_keys = False
+ serialize_eclasses = True
def __init__(self, location, label, auxdbkeys, readonly=False):
""" initialize the derived class; specifically, store label/keys"""
self.readonly = readonly
self.sync_rate = 0
self.updates = 0
-
def __getitem__(self, cpv):
"""set a cpv to values
self.commit()
self.updates = 0
d=self._getitem(cpv)
- if "_eclasses_" in d:
+ if self.serialize_eclasses and "_eclasses_" in d:
d["_eclasses_"] = reconstruct_eclasses(cpv, d["_eclasses_"])
return d
override this in derived classess"""
raise NotImplementedError
-
def __setitem__(self, cpv, values):
"""set a cpv to values
This shouldn't be overriden in derived classes since it handles the readonly checks"""
for k in d.keys():
if d[k] == '':
del d[k]
- if "_eclasses_" in values:
+ if self.serialize_eclasses and "_eclasses_" in values:
d["_eclasses_"] = serialize_eclasses(d["_eclasses_"])
- elif "_eclasses_" in values:
+ elif self.serialize_eclasses and "_eclasses_" in values:
d = ProtectedDict(values)
d["_eclasses_"] = serialize_eclasses(d["_eclasses_"])
else:
self.commit()
self.updates = 0
-
def _setitem(self, name, values):
"""__setitem__ calls this after readonly checks. override it in derived classes
note _eclassees_ key *must* be handled"""
raise NotImplementedError
-
def __delitem__(self, cpv):
"""delete a key from the cache.
This shouldn't be overriden in derived classes since it handles the readonly checks"""
self.commit()
self.updates = 0
-
def _delitem(self,cpv):
"""__delitem__ calls this after readonly checks. override it in derived classes"""
raise NotImplementedError
-
def has_key(self, cpv):
raise NotImplementedError
-
def keys(self):
return tuple(self.iterkeys())
if not self.autocommits:
raise NotImplementedError
+ def __contains__(self, cpv):
+ return self.has_key(cpv)
+
def get_matches(self, match_dict):
"""generic function for walking the entire cache db, matching restrictions to
filter what cpv's are returned. Derived classes should override this if they
cont = False
break
if cont:
-# yield cpv,vals
yield cpv
import elog_modules.mod_save, portage_exec, portage_exception
def process(mysettings, cpv, logentries, fulltext):
- elog_modules.mod_save.process(mysettings, cpv, logentries, fulltext)
+ elogfilename = elog_modules.mod_save.process(mysettings, cpv, logentries, fulltext)
if (not "PORTAGE_ELOG_COMMAND" in mysettings.keys()) \
or len(mysettings["PORTAGE_ELOG_COMMAND"]) == 0:
raise portage_exception.MissingParameter("!!! Custom logging requested but PORTAGE_ELOG_COMMAND is not defined")
else:
mylogcmd = mysettings["PORTAGE_ELOG_COMMAND"]
- mylogcmd.replace("${LOGFILE}", elogfilename)
- mylogcmd.replace("${PACKAGE}", cpv)
+ mylogcmd = mylogcmd.replace("${LOGFILE}", elogfilename)
+ mylogcmd = mylogcmd.replace("${PACKAGE}", cpv)
retval = portage_exec.spawn_bash(mylogcmd)
if retval != 0:
raise portage_exception.PortageException("!!! PORTAGE_ELOG_COMMAND failed with exitcode %d" % retval)
elogfile.write(fulltext)
elogfile.close()
- return
+ return elogfilename
myargs = [os.path.basename(mycmd)]+mysplit[1:]
mypid=os.fork()
if mypid == 0:
- os.execv(mycmd,myargs)
+ try:
+ os.execv(mycmd,myargs)
+ except OSError:
+ pass
sys.stderr.write("!!! Failed to spawn fetcher.\n")
sys.exit(1)
retval=os.waitpid(mypid,0)[1]
signal.signal(signal.SIGTERM, signal.SIG_IGN)
# 0=send to *everybody* in process group
- portageexit()
- print "Exiting due to signal"
- os.kill(0,signum)
sys.exit(1)
signal.signal(signal.SIGCHLD, signal.SIG_DFL)
commands.getstatusoutput("cd / ; "+portage_const.PREFIX+"/sbin/ldconfig -r "+root)
else:
commands.getstatusoutput("cd / ; "+portage_const.PREFIX+"/sbin/ldconfig -X -r "+root)
- elif ostype == "FreeBSD":
+ elif ostype in ("FreeBSD","DragonFly"):
if (ld_cache_update):
writemsg(">>> Regenerating "+str(root)+"var/run/ld-elf.so.hints...\n")
commands.getstatusoutput("cd / ; "+portage_const.PREFIX+"/sbin/ldconfig -i -elf -f "+str(root)+portage_const.PREFIX+"var/run/ld-elf.so.hints "+str(root)+portage_const.PREFIX+"etc/ld.so.conf")
def load_best_module(self,property_string):
best_mod = best_from_dict(property_string,self.modules,self.module_priority)
- return load_mod(best_mod)
+ try:
+ mod = load_mod(best_mod)
+ except:
+ writemsg(red("!!! Failed to import module '%s'\n") % best_mod)
+ sys.exit(1)
+ return mod
def lock(self):
self.locked = 1
if not fetchonly:
fetched=2
else:
- # Check md5sum's at each fetch for fetchonly.
+ # Verify checksums at each fetch for fetchonly.
verified_ok,reason = portage_checksum.verify_all(mysettings["DISTDIR"]+"/"+myfile, mydigests[myfile])
if not verified_ok:
- writemsg("!!! Previously fetched file: "+str(myfile)+"\n!!! Reason: "+reason+"\nRefetching...\n\n")
+ writemsg("!!! Previously fetched file: "+str(myfile)+"\n")
+ writemsg("!!! Reason: "+reason[0]+"\n")
+ writemsg("!!! Got: "+reason[1]+"\n")
+ writemsg("!!! Expected: "+reason[2]+"\n")
+ writemsg("Refetching...\n\n")
os.unlink(mysettings["DISTDIR"]+"/"+myfile)
fetched=0
else:
try:
mystat=os.stat(mysettings["DISTDIR"]+"/"+myfile)
# no exception? file exists. let digestcheck() report
- # an appropriately for size or md5 errors
+ # an appropriately for size or checksum errors
if (mystat[stat.ST_SIZE]<mydigests[myfile]["size"]):
# Fetch failed... Try the next one... Kill 404 files though.
if (mystat[stat.ST_SIZE]<100000) and (len(myfile)>4) and not ((myfile[-5:]==".html") or (myfile[-4:]==".htm")):
fetched=2
break
else:
- # File is the correct size--check the MD5 sum for the fetched
+ # File is the correct size--check the checksums for the fetched
# file NOW, for those users who don't have a stable/continuous
# net connection. This way we have a chance to try to download
# from another mirror...
verified_ok,reason = portage_checksum.verify_all(mysettings["DISTDIR"]+"/"+myfile, mydigests[myfile])
if not verified_ok:
- writemsg("!!! Fetched file: "+str(myfile)+" VERIFY FAILED!\n!!! Reason: "+reason+"\nRemoving corrupt distfile...\n")
+ writemsg("!!! Fetched file: "+str(myfile)+" VERIFY FAILED!\n")
+ writemsg("!!! Reason: "+reason[0]+"\n")
+ writemsg("!!! Got: "+reason[1]+"\n")
+ writemsg("!!! Expected: "+reason[2]+"\n")
+ writemsg("Removing corrupt distfile...\n")
os.unlink(mysettings["DISTDIR"]+"/"+myfile)
fetched=0
else:
print "!!! Given file does not appear to be readable. Does it exist?"
print "!!! File:",myfile
return None
- mydigests[x] = portage_checksum.perform_all(myfile)
+ mydigests[x] = portage_checksum.perform_multiple_checksums(myfile, hashes=portage_const.MANIFEST1_HASH_FUNCTIONS)
mysize = os.stat(myfile)[stat.ST_SIZE]
else:
if x in oldDigest:
myline += " "+mysum
myline += " "+myarchive
myline += " "+str(mysize)
- if sumName != "MD5":
- # XXXXXXXXXXXXXXXX This cannot be used!
- # Older portage make very dumb assumptions about the formats.
- # We need a lead-in period before we break everything.
- continue
mylines.append(myline)
return mylines
def digestParseFile(myfilename):
"""(filename) -- Parses a given file for entries matching:
- MD5 MD5_STRING_OF_HEX_CHARS FILE_NAME FILE_SIZE
- Ignores lines that do not begin with 'MD5' and returns a
- dict with the filenames as keys and [md5,size] as the values."""
+ <checksumkey> <checksum_hex_string> <filename> <filesize>
+ Ignores lines that don't start with a valid checksum identifier
+ and returns a dict with the filenames as keys and {checksumkey:checksum}
+ as the values."""
if not os.path.exists(myfilename):
return None
def digestCheckFiles(myfiles, mydigests, basedir, note="", strict=0):
"""(fileslist, digestdict, basedir) -- Takes a list of files and a dict
of their digests and checks the digests against the indicated files in
- the basedir given. Returns 1 only if all files exist and match the md5s.
+ the basedir given. Returns 1 only if all files exist and match the checksums.
"""
for x in myfiles:
if not mydigests.has_key(x):
print
print red("!!! Digest verification Failed:")
print red("!!!")+" "+str(myfile)
- print red("!!! Reason: ")+reason
+ print red("!!! Reason: ")+reason[0]
+ print red("!!! Got: ")+reason[1]
+ print red("!!! Expected: ")+reason[2]
print
return 0
else:
- print ">>> md5 "+note+" ;-)",x
+ print ">>> checksums "+note+" ;-)",x
return 1
def digestcheck(myfiles, mysettings, strict=0, justmanifest=0):
- """Checks md5sums. Assumes all files have been downloaded."""
+ """Verifies checksums. Assumes all files have been downloaded."""
# archive files
basedir=mysettings["DISTDIR"]+"/"
digestfn=mysettings["FILESDIR"]+"/digest-"+mysettings["PF"]
# Check the portage-related files here.
mymfiles=listdir(pbasedir,recursive=1,filesonly=1,ignorecvs=1,EmptyOnError=1)
manifest_files = mymdigests.keys()
+ # Files unrelated to the build process are ignored for verification by default
for x in ["Manifest", "ChangeLog", "metadata.xml"]:
while x in mymfiles:
mymfiles.remove(x)
"""Calls pkgsplit on a cpv and returns only the cp."""
return pkgsplit(mycpv)[0]
-def dep_parenreduce(mysplit,mypos=0):
- "Accepts a list of strings, and converts '(' and ')' surrounded items to sub-lists"
- while (mypos<len(mysplit)):
- if (mysplit[mypos]=="("):
- firstpos=mypos
- mypos=mypos+1
- while (mypos<len(mysplit)):
- if mysplit[mypos]==")":
- mysplit[firstpos:mypos+1]=[mysplit[firstpos+1:mypos]]
- mypos=firstpos
- break
- elif mysplit[mypos]=="(":
- #recurse
- mysplit=dep_parenreduce(mysplit,mypos=mypos)
- mypos=mypos+1
- mypos=mypos+1
- return mysplit
-
-def dep_opconvert(mysplit,myuse,mysettings):
- "Does dependency operator conversion"
-
- #check_config_instance(mysettings)
-
- mypos=0
- newsplit=[]
- while mypos<len(mysplit):
- if type(mysplit[mypos])==types.ListType:
- newsplit.append(dep_opconvert(mysplit[mypos],myuse,mysettings))
- mypos += 1
- elif mysplit[mypos]==")":
- #mismatched paren, error
- return None
- elif mysplit[mypos]=="||":
- if ((mypos+1)>=len(mysplit)) or (type(mysplit[mypos+1])!=types.ListType):
- # || must be followed by paren'd list
- return None
- try:
- mynew=dep_opconvert(mysplit[mypos+1],myuse,mysettings)
- except SystemExit, e:
- raise
- except Exception, e:
- print "!!! Unable to satisfy OR dependency:",string.join(mysplit," || ")
- raise e
- mynew[0:0]=["||"]
- newsplit.append(mynew)
- mypos += 2
- elif mysplit[mypos][-1]=="?":
- #uses clause, i.e "gnome? ( foo bar )"
- #this is a quick and dirty hack so that repoman can enable all USE vars:
- if (len(myuse)==1) and (myuse[0]=="*") and mysettings:
- # enable it even if it's ! (for repoman) but kill it if it's
- # an arch variable that isn't for this arch. XXX Sparc64?
- k=mysplit[mypos][:-1]
- if k[0]=="!":
- k=k[1:]
- if k not in archlist and k not in mysettings.usemask:
- enabled=1
- elif k in archlist:
- if k==mysettings["ARCH"]:
- if mysplit[mypos][0]=="!":
- enabled=0
- else:
- enabled=1
- elif mysplit[mypos][0]=="!":
- enabled=1
- else:
- enabled=0
- else:
- enabled=0
- else:
- if mysplit[mypos][0]=="!":
- myusevar=mysplit[mypos][1:-1]
- if myusevar in myuse:
- enabled=0
- else:
- enabled=1
- else:
- myusevar=mysplit[mypos][:-1]
- if myusevar in myuse:
- enabled=1
- else:
- enabled=0
- if (mypos+2<len(mysplit)) and (mysplit[mypos+2]==":"):
- #colon mode
- if enabled:
- #choose the first option
- if type(mysplit[mypos+1])==types.ListType:
- newsplit.append(dep_opconvert(mysplit[mypos+1],myuse,mysettings))
- else:
- newsplit.append(mysplit[mypos+1])
- else:
- #choose the alternate option
- if type(mysplit[mypos+1])==types.ListType:
- newsplit.append(dep_opconvert(mysplit[mypos+3],myuse,mysettings))
- else:
- newsplit.append(mysplit[mypos+3])
- mypos += 4
- else:
- #normal use mode
- if enabled:
- if type(mysplit[mypos+1])==types.ListType:
- newsplit.append(dep_opconvert(mysplit[mypos+1],myuse,mysettings))
- else:
- newsplit.append(mysplit[mypos+1])
- #otherwise, continue.
- mypos += 2
- else:
- #normal item
- newsplit.append(mysplit[mypos])
- mypos += 1
- return newsplit
def dep_virtual(mysplit, mysettings):
"Does virtual dependency conversion"
-
-
-
newsplit=[]
for x in mysplit:
if type(x)==types.ListType:
return newsplit
def dep_eval(deplist):
- if len(deplist)==0:
+ if not deplist:
return 1
if deplist[0]=="||":
#or list; we just need one "1"
return 0
return 1
-def dep_zapdeps(unreduced,reduced,vardbapi=None,use_binaries=0):
+def dep_zapdeps(unreduced,reduced,myroot,use_binaries=0):
"""Takes an unreduced and reduced deplist and removes satisfied dependencies.
Returned deplist contains steps that must be taken to satisfy dependencies."""
writemsg("ZapDeps -- %s\n" % (use_binaries), 2)
- if unreduced==[] or unreduced==['||'] :
+ if not reduced or unreduced == ["||"] or dep_eval(reduced):
return []
- if unreduced[0]=="||":
- if dep_eval(reduced):
- #deps satisfied, return empty list.
- return []
+
+ if unreduced[0] != "||":
+ unresolved = []
+ for (dep, satisfied) in zip(unreduced, reduced):
+ if isinstance(dep, list):
+ unresolved += dep_zapdeps(dep, satisfied, myroot, use_binaries=use_binaries)
+ elif not satisfied:
+ unresolved.append(dep)
+ return unresolved
+
+ # We're at a ( || atom ... ) type level
+ deps = unreduced[1:]
+ satisfieds = reduced[1:]
+
+ target = None
+ for (dep, satisfied) in zip(deps, satisfieds):
+ if isinstance(dep, list):
+ atoms = dep_zapdeps(dep, satisfied, myroot, use_binaries=use_binaries)
else:
- #try to find an installed dep.
- ### We use fakedb when --update now, so we can't use local vardbapi here.
- ### This should be fixed in the feature.
- ### see bug 45468.
- ##if vardbapi:
- ## mydbapi=vardbapi
- ##else:
- ## mydbapi=db[root]["vartree"].dbapi
- mydbapi=db[root]["vartree"].dbapi
-
- if db["/"].has_key("porttree"):
- myportapi=db["/"]["porttree"].dbapi
+ atoms = [dep]
+ missing_atoms = [atom for atom in atoms if not db[myroot]["vartree"].dbapi.match(dep_getkey(atom))]
+
+ if not missing_atoms:
+ if isinstance(dep, list):
+ return atoms # Sorted out by the recursed dep_zapdeps call
else:
- myportapi=None
+ target = dep_getkey(dep) # An installed package that's not yet in the graph
+ break
- if use_binaries and db["/"].has_key("bintree"):
- mybinapi=db["/"]["bintree"].dbapi
- writemsg("Using bintree...\n",2)
+ if not target:
+ if use_binaries:
+ missing_atoms = [atom for atom in atoms if not db[myroot]["bintree"].dbapi.match(atom)]
else:
- mybinapi=None
-
- x=1
- candidate=[]
- while x<len(reduced):
- writemsg("x: %s, reduced[x]: %s\n" % (x,reduced[x]), 2)
- if (type(reduced[x])==types.ListType):
- newcand = dep_zapdeps(unreduced[x], reduced[x], vardbapi=vardbapi, use_binaries=use_binaries)
- candidate.append(newcand)
- else:
- if (reduced[x]==False):
- candidate.append([unreduced[x]])
- else:
- candidate.append([])
- x+=1
+ missing_atoms = [atom for atom in atoms if not db[myroot]["porttree"].dbapi.xmatch("match-visible", atom)]
+ if not missing_atoms:
+ target = (dep, satisfied)
+
+ if isinstance(target, tuple): # Nothing matching installed
+ if isinstance(target[0], list): # ... and the first available was a sublist
+ return dep_zapdeps(target[0], target[1], myroot, use_binaries=use_binaries)
+ else: # ... and the first available was a single atom
+ target = dep_getkey(target[0])
+
+ relevant_atoms = [dep for dep in deps if not isinstance(dep, list) and dep_getkey(dep) == target]
+
+ available_pkgs = {}
+ for atom in relevant_atoms:
+ if use_binaries:
+ pkg_list = db["/"]["bintree"].dbapi.match(atom)
+ else:
+ pkg_list = db["/"]["porttree"].dbapi.xmatch("match-visible", atom)
+ if not pkg_list:
+ continue
+ pkg = best(pkg_list)
+ available_pkgs[pkg] = atom
- #use installed and no-masked package(s) in portage.
- for x in candidate:
- match=1
- for pkg in x:
- if not mydbapi.match(pkg):
- match=0
- break
- if myportapi:
- if not myportapi.match(pkg):
- match=0
- break
- if match:
- writemsg("Installed match: %s\n" % (x), 2)
- return x
+ if not available_pkgs:
+ return [unreduced[0]] # All masked
+
+ target_pkg = best(available_pkgs.keys())
+ suitable_atom = available_pkgs[target_pkg]
+ return [suitable_atom]
- # Use binary packages if available.
- if mybinapi:
- for x in candidate:
- match=1
- for pkg in x:
- if not mybinapi.match(pkg):
- match=0
- break
- else:
- writemsg("Binary match: %s\n" % (pkg), 2)
- if match:
- writemsg("Binary match final: %s\n" % (x), 2)
- return x
-
- #use no-masked package(s) in portage tree
- if myportapi:
- for x in candidate:
- match=1
- for pkg in x:
- if not myportapi.match(pkg):
- match=0
- break
- if match:
- writemsg("Porttree match: %s\n" % (x), 2)
- return x
- #none of the no-masked pkg, use the first one
- writemsg("Last resort candidate: %s\n" % (candidate[0]), 2)
- return candidate[0]
- else:
- if dep_eval(reduced):
- #deps satisfied, return empty list.
- return []
- else:
- returnme=[]
- x=0
- while x<len(reduced):
- if type(reduced[x])==types.ListType:
- returnme+=dep_zapdeps(unreduced[x],reduced[x], vardbapi=vardbapi, use_binaries=use_binaries)
- else:
- if reduced[x]==False:
- returnme.append(unreduced[x])
- x += 1
- return returnme
def dep_getkey(mydep):
if not len(mydep):
mydep=mydep[1:]
return mydep
-def cpv_getkey(mycpv):
- myslash=mycpv.split("/")
- mysplit=pkgsplit(myslash[-1])
- mylen=len(myslash)
- if mylen==2:
- return myslash[0]+"/"+mysplit[0]
- elif mylen==1:
- return mysplit[0]
- else:
- return mysplit
-
-def key_expand(mykey,mydb=None,use_cache=1):
- mysplit=mykey.split("/")
- if len(mysplit)==1:
- if mydb and type(mydb)==types.InstanceType:
- for x in settings.categories:
- if mydb.cp_list(x+"/"+mykey,use_cache=use_cache):
- return x+"/"+mykey
- if virts_p.has_key(mykey):
- return(virts_p[mykey][0])
- return "null/"+mykey
- elif mydb:
- if type(mydb)==types.InstanceType:
- if (not mydb.cp_list(mykey,use_cache=use_cache)) and virts and virts.has_key(mykey):
- return virts[mykey][0]
- return mykey
-
-def cpv_expand(mycpv,mydb=None,use_cache=1):
- """Given a string (packagename or virtual) expand it into a valid
- cat/package string. Virtuals use the mydb to determine which provided
- virtual is a valid choice and defaults to the first element when there
- are no installed/available candidates."""
- myslash=mycpv.split("/")
- mysplit=pkgsplit(myslash[-1])
- if len(myslash)>2:
- # this is illegal case.
- mysplit=[]
- mykey=mycpv
- elif len(myslash)==2:
- if mysplit:
- mykey=myslash[0]+"/"+mysplit[0]
- else:
- mykey=mycpv
- if mydb:
- writemsg("mydb.__class__: %s\n" % (mydb.__class__), 1)
- if type(mydb)==types.InstanceType:
- if (not mydb.cp_list(mykey,use_cache=use_cache)) and virts and virts.has_key(mykey):
- writemsg("virts[%s]: %s\n" % (str(mykey),virts[mykey]), 1)
- mykey_orig = mykey[:]
- for vkey in virts[mykey]:
- if mydb.cp_list(vkey,use_cache=use_cache):
- mykey = vkey
- writemsg("virts chosen: %s\n" % (mykey), 1)
- break
- if mykey == mykey_orig:
- mykey=virts[mykey][0]
- writemsg("virts defaulted: %s\n" % (mykey), 1)
- #we only perform virtual expansion if we are passed a dbapi
- else:
- #specific cpv, no category, ie. "foo-1.0"
- if mysplit:
- myp=mysplit[0]
- else:
- # "foo" ?
- myp=mycpv
- mykey=None
- matches=[]
- if mydb:
- for x in settings.categories:
- if mydb.cp_list(x+"/"+myp,use_cache=use_cache):
- matches.append(x+"/"+myp)
- if (len(matches)>1):
- raise ValueError, matches
- elif matches:
- mykey=matches[0]
-
- if not mykey and type(mydb)!=types.ListType:
- if virts_p.has_key(myp):
- mykey=virts_p[myp][0]
- #again, we only perform virtual expansion if we have a dbapi (not a list)
- if not mykey:
- mykey="null/"+myp
- if mysplit:
- if mysplit[2]=="r0":
- return mykey+"-"+mysplit[1]
- else:
- return mykey+"-"+mysplit[1]+"-"+mysplit[2]
- else:
- return mykey
-
def dep_transform(mydep,oldkey,newkey):
origdep=mydep
if not len(mydep):
mydep=mydep[1:]
return prefix+cpv_expand(mydep,mydb=mydb,use_cache=use_cache)+postfix
-def dep_check(depstring,mydbapi,mysettings,use="yes",mode=None,myuse=None,use_cache=1,use_binaries=0):
+def dep_check(depstring,mydbapi,mysettings,use="yes",mode=None,myuse=None,use_cache=1,use_binaries=0,myroot="/"):
"""Takes a depend string and parses the condition."""
#check_config_instance(mysettings)
if myeval:
return [1,[]]
else:
- myzaps = dep_zapdeps(mysplit,mysplit2,vardbapi=mydbapi,use_binaries=use_binaries)
+ myzaps = dep_zapdeps(mysplit,mysplit2,myroot,use_binaries=use_binaries)
mylist = flatten(myzaps)
writemsg("myzaps: %s\n" % (myzaps), 1)
writemsg("mylist: %s\n" % (mylist), 1)
if mydep!=None:
tmp=(len(mydep)>=1)
if deplist[mypos][0]=="!":
- #tmp=not tmp
- # This is ad-hoc code. We should rewrite this later.. (See #52377)
- # The reason is that portage uses fakedb when --update option now.
- # So portage considers that a block package doesn't exist even if it exists.
- # Then, #52377 happens.
- # ==== start
- # emerge checks if it's block or not, so we can always set tmp=False.
- # but it's not clean..
tmp=False
- # ==== end
deplist[mypos]=tmp
else:
#encountered invalid string
mypos=mypos+1
return deplist
+def cpv_getkey(mycpv):
+ myslash=mycpv.split("/")
+ mysplit=pkgsplit(myslash[-1])
+ mylen=len(myslash)
+ if mylen==2:
+ return myslash[0]+"/"+mysplit[0]
+ elif mylen==1:
+ return mysplit[0]
+ else:
+ return mysplit
+
+def key_expand(mykey,mydb=None,use_cache=1):
+ mysplit=mykey.split("/")
+ if len(mysplit)==1:
+ if mydb and type(mydb)==types.InstanceType:
+ for x in settings.categories:
+ if mydb.cp_list(x+"/"+mykey,use_cache=use_cache):
+ return x+"/"+mykey
+ if virts_p.has_key(mykey):
+ return(virts_p[mykey][0])
+ return "null/"+mykey
+ elif mydb:
+ if type(mydb)==types.InstanceType:
+ if (not mydb.cp_list(mykey,use_cache=use_cache)) and virts and virts.has_key(mykey):
+ return virts[mykey][0]
+ return mykey
+
+def cpv_expand(mycpv,mydb=None,use_cache=1):
+ """Given a string (packagename or virtual) expand it into a valid
+ cat/package string. Virtuals use the mydb to determine which provided
+ virtual is a valid choice and defaults to the first element when there
+ are no installed/available candidates."""
+ myslash=mycpv.split("/")
+ mysplit=pkgsplit(myslash[-1])
+ if len(myslash)>2:
+ # this is illegal case.
+ mysplit=[]
+ mykey=mycpv
+ elif len(myslash)==2:
+ if mysplit:
+ mykey=myslash[0]+"/"+mysplit[0]
+ else:
+ mykey=mycpv
+ if mydb:
+ writemsg("mydb.__class__: %s\n" % (mydb.__class__), 1)
+ if type(mydb)==types.InstanceType:
+ if (not mydb.cp_list(mykey,use_cache=use_cache)) and virts and virts.has_key(mykey):
+ writemsg("virts[%s]: %s\n" % (str(mykey),virts[mykey]), 1)
+ mykey_orig = mykey[:]
+ for vkey in virts[mykey]:
+ if mydb.cp_list(vkey,use_cache=use_cache):
+ mykey = vkey
+ writemsg("virts chosen: %s\n" % (mykey), 1)
+ break
+ if mykey == mykey_orig:
+ mykey=virts[mykey][0]
+ writemsg("virts defaulted: %s\n" % (mykey), 1)
+ #we only perform virtual expansion if we are passed a dbapi
+ else:
+ #specific cpv, no category, ie. "foo-1.0"
+ if mysplit:
+ myp=mysplit[0]
+ else:
+ # "foo" ?
+ myp=mycpv
+ mykey=None
+ matches=[]
+ if mydb:
+ for x in settings.categories:
+ if mydb.cp_list(x+"/"+myp,use_cache=use_cache):
+ matches.append(x+"/"+myp)
+ if (len(matches)>1):
+ raise ValueError, matches
+ elif matches:
+ mykey=matches[0]
+
+ if not mykey and type(mydb)!=types.ListType:
+ if virts_p.has_key(myp):
+ mykey=virts_p[myp][0]
+ #again, we only perform virtual expansion if we have a dbapi (not a list)
+ if not mykey:
+ mykey="null/"+myp
+ if mysplit:
+ if mysplit[2]=="r0":
+ return mykey+"-"+mysplit[1]
+ else:
+ return mykey+"-"+mysplit[1]+"-"+mysplit[2]
+ else:
+ return mykey
+
def getmaskingreason(mycpv):
from portage_util import grablines
global portdb
ret=None
if psplit:
for x in self.porttrees:
- # XXX Why are there errors here? XXX
- try:
- file=x+"/"+mysplit[0]+"/"+psplit[0]+"/"+mysplit[1]+".ebuild"
- except SystemExit, e:
- raise
- except Exception, e:
- print
- print "!!! Problem with determining the name/location of an ebuild."
- print "!!! Please report this on IRC and bugs if you are not causing it."
- print "!!! mycpv: ",mycpv
- print "!!! mysplit:",mysplit
- print "!!! psplit: ",psplit
- print "!!! error: ",e
- print
- sys.exit(17)
-
+ file=x+"/"+mysplit[0]+"/"+psplit[0]+"/"+mysplit[1]+".ebuild"
if os.access(file, os.R_OK):
# when found
ret=[file, x]
def getfetchsizes(self,mypkg,useflags=None,debug=0):
# returns a filename:size dictionnary of remaining downloads
mydigest=self.finddigest(mypkg)
- mymd5s=digestParseFile(mydigest)
- if not mymd5s:
+ checksums=digestParseFile(mydigest)
+ if not checksums:
if debug: print "[empty/missing/bad digest]: "+mypkg
return None
filesdict={}
else:
myuris, myfiles = self.getfetchlist(mypkg,useflags=useflags)
#XXX: maybe this should be improved: take partial downloads
- # into account? check md5sums?
+ # into account? check checksums?
for myfile in myfiles:
- if debug and myfile not in mymd5s.keys():
+ if debug and myfile not in checksums.keys():
print "[bad digest]: missing",myfile,"for",mypkg
- elif myfile in mymd5s.keys():
+ elif myfile in checksums.keys():
distfile=settings["DISTDIR"]+"/"+myfile
if not os.access(distfile, os.R_OK):
- filesdict[myfile]=int(mymd5s[myfile]["size"])
+ filesdict[myfile]=int(checksums[myfile]["size"])
return filesdict
def fetch_check(self, mypkg, useflags=None, mysettings=None, all=False):
raise
except:
pass
- getbinpkg.file_get(settings["PORTAGE_BINHOST"]+"/"+tbz2name, mydest, fcmd=settings["RESUMECOMMAND"])
- return
+ return getbinpkg.file_get(settings["PORTAGE_BINHOST"]+"/"+tbz2name, mydest, fcmd=settings["RESUMECOMMAND"])
def getslot(self,mycatpkg):
"Get a slot for a catpkg; assume it exists."
def portageexit():
global uid,portage_gid,portdb,db
if secpass and not os.environ.has_key("SANDBOX_ACTIVE"):
- # wait child process death
- try:
- while True:
- os.wait()
- except OSError:
- #writemsg(">>> All child process are now dead.")
- pass
-
close_portdbapi_caches()
if mtimedb:
# $Id: /var/cvsroot/gentoo-src/portage/pym/portage_checksum.py,v 1.10.2.2 2005/08/10 05:42:03 ferringb Exp $
-from portage_const import PRIVATE_PATH,PRELINK_BINARY
+from portage_const import PRIVATE_PATH,PRELINK_BINARY,HASHING_BLOCKSIZE
import os
import shutil
import stat
import commands
import sha
+
+# actual hash functions first
+
+#dict of all available hash functions
+hashfunc_map = {}
+
+# We _try_ to load this module. If it fails we do the slightly slower fallback.
+try:
+ import fchksum
+
+ def md5hash(filename):
+ return fchksum.fmd5t(filename)
+
+except ImportError:
+ import md5
+ def md5hash(filename):
+ return pyhash(filename, md5)
+hashfunc_map["MD5"] = md5hash
+
+def sha1hash(filename):
+ return pyhash(filename, sha)
+hashfunc_map["SHA1"] = sha1hash
+
+# Keep pycrypto optional for now, there are no internal fallbacks for these
+try:
+ import Crypto.Hash.SHA256
+
+ def sha256hash(filename):
+ return pyhash(filename, Crypto.Hash.SHA256)
+ hashfunc_map["SHA256"] = sha256hash
+except ImportError:
+ pass
+
+try:
+ import Crypto.Hash.RIPEMD
+
+ def rmd160hash(filename):
+ return pyhash(filename, Crypto.Hash.RIPEMD)
+ hashfunc_map["RMD160"] = rmd160hash
+except ImportError:
+ pass
+
+# end actual hash functions
+
prelink_capable = False
if os.path.exists(PRELINK_BINARY):
results = commands.getstatusoutput(PRELINK_BINARY+" --version > /dev/null 2>&1")
def perform_md5(x, calc_prelink=0):
return perform_checksum(x, md5hash, calc_prelink)[0]
-def perform_sha1(x, calc_prelink=0):
- return perform_checksum(x, sha1hash, calc_prelink)[0]
-
def perform_all(x, calc_prelink=0):
mydict = {}
- mydict["SHA1"] = perform_sha1(x, calc_prelink)
- mydict["MD5"] = perform_md5(x, calc_prelink)
+ for k in hashfunc_map.keys():
+ mydict[k] = perform_checksum(x, hashfunc_map[k], calc_prelink)[0]
return mydict
def get_valid_checksum_keys():
- return ["SHA1", "MD5"]
+ return hashfunc_map.keys()
def verify_all(filename, mydict, calc_prelink=0, strict=0):
# Dict relates to single file only.
file_is_ok = True
reason = "Reason unknown"
try:
- if mydict["size"] != os.stat(filename)[stat.ST_SIZE]:
- return False,"Filesize does not match recorded size"
+ mysize = os.stat(filename)[stat.ST_SIZE]
+ if mydict["size"] != mysize:
+ return False,("Filesize does not match recorded size", mysize, mydict["size"])
except OSError, e:
return False, str(e)
for x in mydict.keys():
if x == "size":
continue
- elif x == "SHA1":
- if mydict[x] != perform_sha1(filename, calc_prelink=calc_prelink):
+ elif x in hashfunc_map.keys():
+ myhash = perform_checksum(filename, hashfunc_map[x], calc_prelink=calc_prelink)[0]
+ if mydict[x] != myhash:
if strict:
raise portage_exception.DigestException, "Failed to verify '$(file)s' on checksum type '%(type)s'" % {"file":filename, "type":x}
else:
file_is_ok = False
- reason = "Failed on %s verification" % (x,)
- break
- elif x == "MD5":
- if mydict[x] != perform_md5(filename, calc_prelink=calc_prelink):
- if strict:
- raise portage_exception.DigestException, "Failed to verify '$(file)s' on checksum type '%(type)s'" % {"file":filename, "type":x}
- else:
- file_is_ok = False
- reason = "Failed on %s verification" % (x,)
+ reason = (("Failed on %s verification" % x), myhash,mydict[x])
break
return file_is_ok,reason
-# We _try_ to load this module. If it fails we do the slow fallback.
-try:
- import fchksum
-
- def md5hash(filename):
- return fchksum.fmd5t(filename)
-
-except ImportError:
- import md5
- def md5hash(filename):
- f = open(filename, 'rb')
- blocksize=32768
- data = f.read(blocksize)
- size = 0L
- sum = md5.new()
- while data:
- sum.update(data)
- size = size + len(data)
- data = f.read(blocksize)
- f.close()
-
- return (sum.hexdigest(),size)
-
-def sha1hash(filename):
+def pyhash(filename, hashobject):
f = open(filename, 'rb')
- blocksize=32768
+ blocksize = HASHING_BLOCKSIZE
data = f.read(blocksize)
size = 0L
- sum = sha.new()
+ sum = hashobject.new()
while data:
sum.update(data)
size = size + len(data)
data = f.read(blocksize)
f.close()
- return (sum.hexdigest(),size)
+ return (sum.hexdigest(), size)
def perform_checksum(filename, hash_function=md5hash, calc_prelink=0):
myfilename = filename[:]
portage_locks.unlockfile(mylock)
return (myhash,mysize)
+
+def perform_multiple_checksums(filename, hashes=["MD5"], calc_prelink=0):
+ rVal = {}
+ for x in hashes:
+ rVal[x] = perform_checksum(filename, hashfunc_map[x], calc_prelink)[0]
+ return rVal
EAPI = "prefix"
+HASHING_BLOCKSIZE = 32768
+MANIFEST1_HASH_FUNCTIONS = ["MD5","SHA256","RMD160"]
+
# ===========================================================================
# END OF CONSTANTS -- END OF CONSTANTS -- END OF CONSTANTS -- END OF CONSTANT
# ===========================================================================
# all vars that are to wind up in portage_const must have their name listed in __all__
-__all__ = ["PREFIX", "SYSCONFDIR", "PORTAGE_BASE", "wheelgid"]
+__all__ = ["PREFIX", "SYSCONFDIR", "PORTAGE_BASE", "portageuser", "portagegroup", "rootuser", "wheelgid", "wheelgroup"]
PREFIX="@DOMAIN_PREFIX@/"
SYSCONFDIR="@sysconfdir@/"
PORTAGE_BASE="@PORTAGE_BASE@/"
+portagegroup="@portagegroup@"
+portageuser="@portageuser@"
+rootuser="@rootuser@"
wheelgid="@wheelgid@"
+wheelgroup"@wheelgroup@
# Quick validity checks
for x in range(len(deparray)):
if deparray[x] in ["||","&&"]:
- if len(deparray) == x:
+ if len(deparray) - 1 == x:
# Operator is the last element
raise portage_exception.InvalidDependString("INVALID "+deparray[x]+" DEPEND STRING: "+str(deparray))
if type(deparray[x+1]) != types.ListType:
projects / portage.git / commitdiff