Zac Medico [Wed, 12 Jan 2011 15:30:00 +0000 (07:30 -0800)]
docbook: Add missing "of" from last commit.
Zac Medico [Wed, 12 Jan 2011 15:15:20 +0000 (07:15 -0800)]
Don't filter QA warning for --disable-dep...
This reverts commit
f3785f651bff9febc0bb3be4f0ee8e4d732495d5.
It's unnecessary to filter since now we check configure --help output
for disable-dependency-tracking (bug #211529).
Zac Medico [Wed, 12 Jan 2011 15:13:14 +0000 (07:13 -0800)]
econf: check confgure --help for --disable-dep...
This is the new behavior, for bug #211529.
Zac Medico [Mon, 10 Jan 2011 05:31:55 +0000 (21:31 -0800)]
ebuild.sh: skip QA_* init in "depend" phase
Zac Medico [Fri, 7 Jan 2011 00:34:17 +0000 (16:34 -0800)]
Scheduler: free unneeded uninstall task from mem
Zac Medico [Thu, 6 Jan 2011 04:17:03 +0000 (20:17 -0800)]
Scheduler: handle same cpv with different SLOT
Zac Medico [Thu, 6 Jan 2011 03:51:58 +0000 (19:51 -0800)]
Scheduler: fix reposyntax uninstall hash breakage
Zac Medico [Thu, 6 Jan 2011 03:32:32 +0000 (19:32 -0800)]
Scheduler: always share FakeVartree pkg_cache
Zac Medico [Thu, 6 Jan 2011 03:12:45 +0000 (19:12 -0800)]
Scheduler: avoid unecessary Package instances
Zac Medico [Thu, 6 Jan 2011 01:58:03 +0000 (17:58 -0800)]
Scheduler: cache more Packages in --nodeps mode
Zac Medico [Thu, 6 Jan 2011 01:41:28 +0000 (17:41 -0800)]
Scheduler: cache Package instance more
This helps to avoid some cases in which emerge crashes due to ebuild
modification that leads to failure in aux_get/manifest check while
the Scheduler is running. Typically such modifications are harmless,
so it's beneficial to cache the metadata so that aux_get is less
likely to be called when it will fail.
Mike Frysinger [Wed, 5 Jan 2011 06:55:54 +0000 (01:55 -0500)]
emerge-webrsync: eat trailing newlines
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Mike Frysinger [Wed, 5 Jan 2011 06:55:28 +0000 (01:55 -0500)]
emerge-webrsync: add xz support
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Zac Medico [Wed, 5 Jan 2011 04:03:51 +0000 (20:03 -0800)]
REQUIRED_USE: when unsatisfied, show current USE
Zac Medico [Wed, 5 Jan 2011 03:52:36 +0000 (19:52 -0800)]
Split out pkg_use_display() from action_info().
Zac Medico [Wed, 5 Jan 2011 01:26:22 +0000 (17:26 -0800)]
REQUIRED_USE: test unsatisfied without USE dep
Zac Medico [Wed, 5 Jan 2011 00:52:32 +0000 (16:52 -0800)]
composite_db: use ebuild visibility for installed
This will fix bug #350488.
Zac Medico [Tue, 4 Jan 2011 23:47:54 +0000 (15:47 -0800)]
dep_zapdeps: add other_installed_some choice bin
Zac Medico [Tue, 4 Jan 2011 01:43:40 +0000 (17:43 -0800)]
Zac Medico [Tue, 4 Jan 2011 01:13:24 +0000 (17:13 -0800)]
dep_zapdeps: add other_installed choice bin
Hopefully this will fix bug #350488.
Zac Medico [Tue, 4 Jan 2011 00:37:39 +0000 (16:37 -0800)]
Fix spelling of unsatisfied from previous commit.
Zac Medico [Tue, 4 Jan 2011 00:13:41 +0000 (16:13 -0800)]
REQUIRED_USE: always show requirements
This fixes a case in which the user would receive a "masked by:
REQUIRED_USE violated" message without any more information about
the specific requirements. Now, unsatisfied REQUIRED_USE is never
treated as a mask, since it really is nothing like a mask. It's much
more like an unsatisfied USE dependency, so we need to handle it in
a similar way.
Zac Medico [Mon, 3 Jan 2011 09:01:54 +0000 (01:01 -0800)]
depgraph: merge some logic from bug #350254
Zac Medico [Mon, 3 Jan 2011 00:15:45 +0000 (16:15 -0800)]
create_trees: whitelist SSH_AGENT_PID/AUTH_SOCK
This variables can be used for fetching from binhost.
Zac Medico [Mon, 3 Jan 2011 00:02:09 +0000 (16:02 -0800)]
add depclean test for bug #350285
Zac Medico [Sun, 2 Jan 2011 23:52:01 +0000 (15:52 -0800)]
doc use_with/enable empty 3rd arg in EAPI 4
See bug #322049.
Zac Medico [Sun, 2 Jan 2011 23:13:46 +0000 (15:13 -0800)]
DepcleanWithInstalledMasked: fix breakage
An assumption about KEYWORDS masking behavior for installed packages
was broken by commit
b7de3d1fb10e40b4ef03462f8a145968689ac587.
Zac Medico [Sun, 2 Jan 2011 22:54:00 +0000 (14:54 -0800)]
Try to avoid downgrade due to unsat REQUIRED_USE.
Zac Medico [Sun, 2 Jan 2011 22:45:58 +0000 (14:45 -0800)]
mod_syslog: call syslog() once per line
This will fix bug #350379.
Zac Medico [Sun, 2 Jan 2011 21:10:42 +0000 (13:10 -0800)]
depclean: account for more masks
Account for packages with masks (like KEYWORDS masks) that are usually
ignored in visibility checks for installed packages, in order to
handle cases like bug #350285.
Ulrich Mueller [Sun, 2 Jan 2011 16:28:48 +0000 (17:28 +0100)]
newins in EAPI 4 should actually install the file. Fixes bug 350385.
Zac Medico [Sun, 2 Jan 2011 10:20:46 +0000 (02:20 -0800)]
MetaDataXML: fix broken upstream initialization
Zac Medico [Sun, 2 Jan 2011 10:01:43 +0000 (02:01 -0800)]
depgraph: use Package.masks for installed
In some cases we do special masking checks for installed packages,
and Package.masks is a convenient way to do it.
Zac Medico [Sun, 2 Jan 2011 09:52:02 +0000 (01:52 -0800)]
depclean: account for more masks
Account for packages with masks (like KEYWORDS masks) that are usually
ignored in visibility checks for installed packages, in order to
handle cases like bug #350285.
Ulrich Mueller [Sat, 1 Jan 2011 16:08:44 +0000 (17:08 +0100)]
Filter QA warning for --disable-dependency-tracking configure option.
Ulrich Mueller [Sat, 1 Jan 2011 19:37:42 +0000 (20:37 +0100)]
Update copyright date.
Arfrever Frehtes Taifersar Arahesis [Sat, 1 Jan 2011 15:49:38 +0000 (16:49 +0100)]
Update comments.
Zac Medico [Sat, 1 Jan 2011 04:38:08 +0000 (20:38 -0800)]
Deprecate EAPI 4_pre1 and drop 2_pre* support.
Zac Medico [Sat, 1 Jan 2011 03:09:11 +0000 (19:09 -0800)]
egencache: handle ImportError for ExpatError
Zac Medico [Fri, 31 Dec 2010 21:50:57 +0000 (13:50 -0800)]
NEWS: EAPI 4 support in portage-2.1.9.27
Zac Medico [Fri, 31 Dec 2010 20:17:59 +0000 (12:17 -0800)]
dyn_clean: remove .pretended
Zac Medico [Fri, 31 Dec 2010 11:27:03 +0000 (03:27 -0800)]
Omit user:pass@ from binhost Packages cache path.
Zac Medico [Fri, 31 Dec 2010 08:51:24 +0000 (00:51 -0800)]
bintree: add timeout for closing binhost conn
Hopefully this solves bug #350139.
Zac Medico [Fri, 31 Dec 2010 08:12:03 +0000 (00:12 -0800)]
mkrelease.sh: set portage.VERSION
Zac Medico [Fri, 31 Dec 2010 07:55:19 +0000 (23:55 -0800)]
Binpkg: implement REPLACING_VERSIONS
Zac Medico [Fri, 31 Dec 2010 07:12:16 +0000 (23:12 -0800)]
ebuild(5): merge EAPI 4 docs from master
Zac Medico [Fri, 31 Dec 2010 07:10:36 +0000 (23:10 -0800)]
tests: emerge EAPI 4 from master
Zac Medico [Thu, 30 Dec 2010 22:23:57 +0000 (14:23 -0800)]
portage.const: bump EAPI to 4 (council approved)
Zac Medico [Thu, 30 Dec 2010 22:14:47 +0000 (14:14 -0800)]
docbook: EAPI 4_pre1 is 4 (council approved)
Zac Medico [Fri, 31 Dec 2010 06:56:34 +0000 (22:56 -0800)]
Merge dep_getrepo test from master.
Zac Medico [Wed, 29 Dec 2010 19:04:02 +0000 (11:04 -0800)]
fetch: don't require g+s mode on $DISTDIR
The g+s mode doesn't seem to help anyone (including live ebuilds
since they typically don't do recursive g+w). We already do manual
application of portage group privileges to individual files, so the
g+s mode is redundant.
Zac Medico [Wed, 29 Dec 2010 18:51:28 +0000 (10:51 -0800)]
unpack: handle echo failure in unzip pipe
This will fix bug #338744.
Zac Medico [Wed, 29 Dec 2010 18:28:25 +0000 (10:28 -0800)]
bash: as a general rule, use read -r
This solves cases like bug #350045.
Ulrich Müller [Wed, 29 Dec 2010 18:11:21 +0000 (10:11 -0800)]
doins: use read -r
This will fix bug #350045.
Michał Górny [Wed, 29 Dec 2010 13:37:25 +0000 (14:37 +0100)]
EbuildPhase: Display upstream info before setup.
Display upstream maintainers and bugtrackers along with Gentoo
maintainers when emerging a package.
Zac Medico [Tue, 28 Dec 2010 21:58:16 +0000 (13:58 -0800)]
fetch: fix listonly/allow_missing_digests logic
Zac Medico [Thu, 23 Dec 2010 19:40:55 +0000 (11:40 -0800)]
_exec: disable SIGQUIT handler override by parent
This will fix bug #289486.
Zac Medico [Tue, 28 Dec 2010 21:44:57 +0000 (13:44 -0800)]
EbuildFetcher: bail out if missing digest
This will fix bug #348864.
Zac Medico [Tue, 28 Dec 2010 21:12:30 +0000 (13:12 -0800)]
Use EbuildFetcher for --fetchonly (not --pretend)
Zac Medico [Tue, 28 Dec 2010 20:41:26 +0000 (12:41 -0800)]
EbuildFetcher: pass digests to fetch()
Zac Medico [Tue, 28 Dec 2010 17:07:40 +0000 (09:07 -0800)]
doebuild: validate more Package metadata keys
Zac Medico [Mon, 27 Dec 2010 22:47:28 +0000 (14:47 -0800)]
unmerge: ignore rmdir EPERM from mount point
This will fix bug #349757.
Zac Medico [Mon, 27 Dec 2010 21:40:45 +0000 (13:40 -0800)]
ebuild(1): use Package to validate metadata
Zac Medico [Sun, 26 Dec 2010 21:22:38 +0000 (13:22 -0800)]
anydbm: use gdbm "u" flag for concurrent writers
This will fix bug #53607.
Zac Medico [Sat, 25 Dec 2010 00:46:01 +0000 (16:46 -0800)]
depgraph: Optimize --newuse with Package attrs.
Zac Medico [Wed, 22 Dec 2010 22:15:18 +0000 (14:15 -0800)]
Fix portage restart warning message.
Zac Medico [Wed, 22 Dec 2010 19:35:29 +0000 (11:35 -0800)]
Package.operation: eliminate duplicate code
Zac Medico [Tue, 21 Dec 2010 18:02:18 +0000 (10:02 -0800)]
doebuild: remove unused emerge/returnpid logic
Zac Medico [Tue, 21 Dec 2010 17:35:25 +0000 (09:35 -0800)]
Display relevant FEATURES before setup phase
Thanks to Christian Ruppert <idl0r@gentoo.org> for the suggestion.
Zac Medico [Tue, 21 Dec 2010 17:05:01 +0000 (09:05 -0800)]
doebuild: fix more bug #349273
Zac Medico [Tue, 21 Dec 2010 16:29:39 +0000 (08:29 -0800)]
doebuild: fix bug #349273
Since commit
75d3130fc8293fff68389b8fed44b8d47a0abe80, doebuild()
for the 'setup' phase triggers the actionmap_deps code path in order
to call the 'pretend' phase as a dependency. This path previously
assumed a portdbapi instance, which may not hold for some api consumers
such as entropy. Therefore, fix the code path to work with other dbapi
types.
Zac Medico [Mon, 20 Dec 2010 15:53:43 +0000 (07:53 -0800)]
saveNomergeFavorites: remove unecessary vars
Zac Medico [Mon, 20 Dec 2010 15:47:48 +0000 (07:47 -0800)]
saveNomergeFavorites: fix possible root confusion
Zac Medico [Mon, 20 Dec 2010 13:48:39 +0000 (05:48 -0800)]
Use write_atomic to rewrite lafiles.
This should succeed even in some cases in which a normal write might
fail due to file permission settings on some operating systems such as
HP-UX.
Zac Medico [Mon, 20 Dec 2010 03:33:26 +0000 (19:33 -0800)]
Substitute bash builtin >> for touch command.
It should be faster to use the bash builtin, and also allows us to
avoid issues with touch breakage (like in bug #348640).
Zac Medico [Sun, 19 Dec 2010 18:12:42 +0000 (10:12 -0800)]
REQUIRED_USE: describe ^^ as exactly-one-of
Zac Medico [Sun, 19 Dec 2010 17:57:18 +0000 (09:57 -0800)]
paren_enclose: preserve atom USE conditionals
The depgraph sometimes uses paren_enclose to temporarily convert parts
of use_reduce output to a plain string. In these cases, we want to
preserve atom USE conditionals, since these sometimes make a difference
for IUSE checks during package matching (as in bug #348389).
Zac Medico [Sat, 18 Dec 2010 15:01:16 +0000 (07:01 -0800)]
repoman: ignore deprecated profiles.desc entries
Zac Medico [Fri, 17 Dec 2010 21:08:16 +0000 (13:08 -0800)]
digestcheck: avoid display of u'' from KeyError
In python-2.x, KeyError.__str__ produces unicode literals that we
don't want users to see.
Zac Medico [Fri, 17 Dec 2010 20:12:54 +0000 (12:12 -0800)]
repoman: handle ElementTree.ParseError
Zac Medico [Thu, 16 Dec 2010 20:29:22 +0000 (12:29 -0800)]
getbinpkg.py: auto-handle NotImplementedError
This should have been done as part of bug #338190.
Zac Medico [Thu, 16 Dec 2010 07:24:16 +0000 (23:24 -0800)]
getbinpkg.py: use local HTTPSConnection import
Use local import since https typically isn't needed, and this way we
can usually avoid triggering the global scope http.client ImportError
handler (like during stage1 -> stage2 builds where USE=ssl is disabled
for python).
Zac Medico [Thu, 16 Dec 2010 02:55:53 +0000 (18:55 -0800)]
slot_conflict_handler: fix for bug #348389
This fixes a broken assumption about USE dep matching behavior,
since commit
766988aa56bb1bbd2796f96946e62ed78cdf259 introduced
stricter IUSE matching by using Atom.unevaluated_atom for IUSE
checks (thus changing behavior in cases where conditional USE
dependencies disappear upon evaluation).
Zac Medico [Thu, 16 Dec 2010 02:41:24 +0000 (18:41 -0800)]
Atom.without_use: handle unevaluated_atom.use
Atom.unevaluated_atom.use is used for IUSE checks when matching
packages, so it must not propagate to Atom.without_use.
Zac Medico [Thu, 16 Dec 2010 01:21:13 +0000 (17:21 -0800)]
highlight_violations: handle unversioned atom
A test case is included, which triggers the previously broken case.
Fabian Groffen [Wed, 15 Dec 2010 17:58:45 +0000 (18:58 +0100)]
removeFromContents: don't use EROOT
Fix bug #348790, make sure that CONTENTS entries keep their EPREFIX, by
using ROOT, instead of EROOT.
Zac Medico [Tue, 14 Dec 2010 04:29:10 +0000 (20:29 -0800)]
etc-update: eliminate duplicate example configs
Zac Medico [Tue, 14 Dec 2010 03:47:27 +0000 (19:47 -0800)]
apply_recursive_permissions: don't follow symlink
This will fix the following error, reported in bug #339670:
_ebuild_exit
_post_phase_userpriv_perms(settings)
File "/usr/lib64/portage/pym/portage/package/ebuild/doebuild.py", line 1207,
in _post_phase_userpriv_perms
filemode=0o60, filemask=0)
File "/usr/lib64/portage/pym/portage/util/__init__.py", line 1018, in
apply_recursive_permissions
uid=uid, gid=gid, mode=filemode, mask=filemask)
File "/usr/lib64/portage/pym/portage/util/__init__.py", line 1043, in
apply_secpass_permissions
stat_cached = os.stat(filename)
File "/usr/lib64/portage/pym/portage/__init__.py", line 226, in __call__
rval = self._func(*wrapped_args, **wrapped_kwargs)
OSError: [Errno 40] Too many levels of symbolic links:
'/var/tmp/portage/dev-vcs/mercurial-1.7.2/temp/tests-2.7/child0/test-mq-symlinks.t/linka'
Anthony Basile [Tue, 14 Dec 2010 01:35:57 +0000 (20:35 -0500)]
etc-update: tweak style in do_distconf()
Zac Medico [Mon, 13 Dec 2010 06:22:53 +0000 (22:22 -0800)]
etc-update: replace 'seq' with pure bash
Anthony Basile [Sun, 12 Dec 2010 23:30:45 +0000 (18:30 -0500)]
etc-update: add option to save example config
This will fix bug #348561.
Zac Medico [Sun, 12 Dec 2010 12:36:20 +0000 (04:36 -0800)]
xtermTitleReset: revert behavior for bug #348460
Zac Medico [Sun, 12 Dec 2010 10:21:50 +0000 (02:21 -0800)]
BinpkgVerifier: don't unlink emerge-fetch.log
Zac Medico [Wed, 8 Dec 2010 19:23:48 +0000 (11:23 -0800)]
Preserve existing hardlinks during merge.
The previous code created hardlinks rather aggressively, which was
helpful as a workaround for lack of hardlinks in tarballs created
by quickpkg due to bug #185305. Since bug #338509, quickpkg tarballs
preserve hardlinks, so there's no need for aggressive hardlink
creation. Therefore, simply preserve existing hardlinks during merge,
by comparison of st_dev and st_ino from the source files.
Zac Medico [Wed, 8 Dec 2010 05:25:23 +0000 (21:25 -0800)]
quickpkg: revert hardlink fix from bug #185305
For some reason, bug #185305 no longer seems to be an issue, and the
fix/workaround seems to cause bug 338509.
Mike Frysinger [Wed, 8 Dec 2010 04:17:38 +0000 (23:17 -0500)]
quickpkg: may -> might reword
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Arfrever Frehtes Taifersar Arahesis [Mon, 6 Dec 2010 12:50:52 +0000 (13:50 +0100)]
Use bytes literals.
Zac Medico [Mon, 6 Dec 2010 00:56:04 +0000 (16:56 -0800)]
inject binpkg after fetch for pkg_pretend
Zac Medico [Mon, 6 Dec 2010 00:07:48 +0000 (16:07 -0800)]
Fix pkg_pretend output and error handling.
Zac Medico [Sun, 5 Dec 2010 23:15:09 +0000 (15:15 -0800)]
Scheduler: Always display message for pkg_pretend