Zac Medico [Wed, 12 Sep 2012 06:24:44 +0000 (23:24 -0700)]
Replace getstatusoutput with unicode safe Popen.
This fixes potential issues similar to those reported in bug #310789.
Zac Medico [Wed, 12 Sep 2012 04:56:39 +0000 (21:56 -0700)]
make_herd_base: encode unicode file paths safely
This will fix cases similar to the one fixed in commit
f01d7a8ca158ec60df76a3e8ae5b80ac3f62429e for bug #310789.
Zac Medico [Wed, 12 Sep 2012 04:52:33 +0000 (21:52 -0700)]
egencache: encode unicode file paths safely
This will fix cases similar to the one fixed in commit
f01d7a8ca158ec60df76a3e8ae5b80ac3f62429e for bug #310789.
Zac Medico [Wed, 12 Sep 2012 04:46:43 +0000 (21:46 -0700)]
MetaDataXML: encode unicode file paths safely
This will fix cases similar to the one fixed in commit
f01d7a8ca158ec60df76a3e8ae5b80ac3f62429e for bug #310789.
Zac Medico [Wed, 12 Sep 2012 04:26:49 +0000 (21:26 -0700)]
Add EAPI 5_pre2 with features approved by council
In the 9/11 council meeting, they approved everything in EAPI 5_pre1
except user patches. So, add EAPI 5_pre2 which is identical to EAPI
5_pre1 but with user patches (apply_user_patches) support removed.
Michael Palimaka [Tue, 11 Sep 2012 17:22:05 +0000 (03:22 +1000)]
Only validate .desktop files that are installed into XDG-compliant locations.
This replaces the existing error blacklist, and therefore removes the repoman check.
Zac Medico [Mon, 10 Sep 2012 21:06:52 +0000 (14:06 -0700)]
repoman: fix Unicode unequal comparison warning
Zac Medico [Mon, 10 Sep 2012 20:45:23 +0000 (13:45 -0700)]
repoman: fix os.system() unicode, bug #310789
Zac Medico [Mon, 10 Sep 2012 20:33:35 +0000 (13:33 -0700)]
repoman: ElementTree.parse: encode unicode path
This fixes another issue like bug #310789.
Zac Medico [Mon, 10 Sep 2012 19:46:50 +0000 (12:46 -0700)]
use(): use set -f and handle IFS
Zac Medico [Mon, 10 Sep 2012 16:40:22 +0000 (09:40 -0700)]
Improve ignored binpkg display for bug #434540.
Zac Medico [Mon, 10 Sep 2012 05:45:54 +0000 (22:45 -0700)]
repoman: fix getstatusoutput more, bug #310789
Zac Medico [Mon, 10 Sep 2012 01:26:19 +0000 (18:26 -0700)]
repoman: fix getstatusoutput unicode, bug #310789
Zac Medico [Mon, 10 Sep 2012 00:53:22 +0000 (17:53 -0700)]
repoman: fix popen unicode handling, bug #310789
Zac Medico [Mon, 10 Sep 2012 00:07:18 +0000 (17:07 -0700)]
ObjectProxy: implement __enter__ and __exit__
Joachim Bartosik [Sun, 9 Sep 2012 04:16:36 +0000 (21:16 -0700)]
DEVELOPING: fix errors for bug #313413
Zac Medico [Sun, 9 Sep 2012 01:07:16 +0000 (18:07 -0700)]
EbuildPhase: simplify utime args
Zac Medico [Sun, 9 Sep 2012 00:54:10 +0000 (17:54 -0700)]
EbuildPhase: bump WORKDIR timestamp after unpack
This will fix bug #332217.
Zac Medico [Sat, 8 Sep 2012 20:31:50 +0000 (13:31 -0700)]
egencache: tweak metadata-transfer logic
Now is won't force metadata-transfer unless it's absolutely necessary.
Zac Medico [Sat, 8 Sep 2012 16:50:19 +0000 (09:50 -0700)]
egencache: skip metadata-transfer more often
Zac Medico [Sat, 8 Sep 2012 16:15:39 +0000 (09:15 -0700)]
best/has_version: --host-root first arg only
Zac Medico [Sat, 8 Sep 2012 05:35:03 +0000 (22:35 -0700)]
egencache: skip metadata-transfer if possible
If there is an existing metadata/md5-cache directory, then it's not
necessary to forcibly enable metadata-transfer.
Zac Medico [Sat, 8 Sep 2012 04:32:03 +0000 (21:32 -0700)]
Detect md5-cache when no cache-formats specified.
Auto-detect cache-formats from the corresponding directories when
cache-formats is not specifed in layout.conf, and prefer md5-dict if
available. After this behavior is deployed in stable portage, the
default egencache format can be changed to md5-dict.
Zac Medico [Sat, 8 Sep 2012 04:07:45 +0000 (21:07 -0700)]
Test portdbapi/egencache cache-formats behavior.
Zac Medico [Fri, 7 Sep 2012 20:52:03 +0000 (13:52 -0700)]
UseManager: add compat code for getUseForce/PUSE
Similar code was added to getUseMask in commit
af50fee789e71c1fff0b405db4cf55b618b73466.
Arfrever Frehtes Taifersar Arahesis [Fri, 7 Sep 2012 00:46:07 +0000 (02:46 +0200)]
Support repository-specific use.stable.force, use.stable.mask,
package.use.stable.force and package.use.stable.mask in EAPI >=5.
Arfrever Frehtes Taifersar Arahesis [Thu, 6 Sep 2012 21:45:26 +0000 (23:45 +0200)]
Update comment.
Zac Medico [Thu, 6 Sep 2012 20:06:12 +0000 (13:06 -0700)]
dispatch-conf.conf: say install rcs, bug #380551
Martin Väth [Wed, 5 Sep 2012 20:36:43 +0000 (13:36 -0700)]
DateSet: use BUILD_TIME, bug #433964
Zac Medico [Tue, 4 Sep 2012 20:40:21 +0000 (13:40 -0700)]
Enable @x11-module-rebuild for 2.1 branch.
Zac Medico [Tue, 4 Sep 2012 20:25:28 +0000 (13:25 -0700)]
Enable @live-rebuild for 2.1 branch.
Zac Medico [Tue, 4 Sep 2012 03:28:25 +0000 (20:28 -0700)]
Deprecate @installed for bug #387059.
Zac Medico [Tue, 4 Sep 2012 03:07:29 +0000 (20:07 -0700)]
EbuildBuildDir: ignore rmdir failure, bug #400641
Zac Medico [Tue, 4 Sep 2012 01:33:26 +0000 (18:33 -0700)]
etc-update: add --quiet for bug #416917
Zac Medico [Tue, 4 Sep 2012 01:05:56 +0000 (18:05 -0700)]
ebuild.5: clarify pkg_nofetch for bug #423403
Michael Palimaka [Mon, 3 Sep 2012 15:08:36 +0000 (01:08 +1000)]
validate_desktop_entry: filter more KDE noise.
Zac Medico [Sun, 2 Sep 2012 23:46:54 +0000 (16:46 -0700)]
emaint cleanconfmem: fix writedict breakage
Use grabdict so the values are stored as lists like writedict expects.
Zac Medico [Sun, 2 Sep 2012 23:03:14 +0000 (16:03 -0700)]
Rename _slot_abi.py to _slot_operator.py.
Zac Medico [Sun, 2 Sep 2012 22:56:46 +0000 (15:56 -0700)]
Quote ${EAPI} more.
This will avoid a potential syntax error in save-ebuild-env.sh if a
saved environment containing corrupt EAPI is sourced.
Zac Medico [Sun, 2 Sep 2012 22:31:20 +0000 (15:31 -0700)]
DateSet: only allow one format option in config
Zac Medico [Sun, 2 Sep 2012 22:13:24 +0000 (15:13 -0700)]
ebuild-helpers/bsd/sed: whitespace + quoting
Zac Medico [Sun, 2 Sep 2012 21:56:22 +0000 (14:56 -0700)]
Unconditionally insert our pym dir in sys.path.
This is especially necessary when running in prefix mode, since
sys.path may contain the path for a different version of portage.
Zac Medico [Sun, 2 Sep 2012 20:33:06 +0000 (13:33 -0700)]
emaint cleanconfmem: handle ROOT
Martin Väth [Sun, 2 Sep 2012 20:18:20 +0000 (13:18 -0700)]
Add DateSet, bug #433704.
Allow sets of packages installed before or after a specific date.
The date may be specified explicitly (in dateformat which defaults to "%x %X")
or seconds since Epoch, or implicitly as an installation date or a filestamp.
For example, the following sets can be specified in /etc/portage/sets.conf:
[date1]
class = portage.sets.dbapi.DateSet
mode = older
date = 08/31/12 10:00:00
dateformat = %%x %%X
[date2]
class = portage.sets.dbapi.DateSet
mode = newer
seconds =
1346400000
[date3]
class = portage.sets.dbapi.DateSet
package = sys-devel/gcc:4.7
[date4]
class = portage.sets.dbapi.DateSet
filestamp = /usr/bin/gcc
Zac Medico [Sun, 2 Sep 2012 20:04:38 +0000 (13:04 -0700)]
emaint cleanconfmem: use writedict
Zac Medico [Sun, 2 Sep 2012 19:46:46 +0000 (12:46 -0700)]
emaint cleanconfmem: ignore md5, bug #433694
Zac Medico [Sun, 2 Sep 2012 19:31:23 +0000 (12:31 -0700)]
validate_desktop_entry: handle emacs.desktop
See <https://bugs.freedesktop.org/show_bug.cgi?id=35844#c6>.
Zac Medico [Sun, 2 Sep 2012 02:37:56 +0000 (19:37 -0700)]
Enable EAPI 5_pre1 for testing.
Zac Medico [Sun, 2 Sep 2012 02:37:15 +0000 (19:37 -0700)]
Document EAPI 5_pre1.
Zac Medico [Sun, 2 Sep 2012 02:24:46 +0000 (19:24 -0700)]
apply_user_patches: die if EBUILD_PHASE is wrong
Zac Medico [Sun, 2 Sep 2012 00:42:11 +0000 (17:42 -0700)]
glsa.py: python3: ResourceWarning: unclosed file
Zac Medico [Sat, 1 Sep 2012 23:08:16 +0000 (16:08 -0700)]
ebuild-helpers/sed: move to bsd subdirectory
This helps to prevent inappropriate use of the sed wrapper on GNU
systems, which commonly occurs when users install portage manually
during recovery situations.
Zac Medico [Sat, 1 Sep 2012 22:51:25 +0000 (15:51 -0700)]
ebuild-helpers/sed: use -ef to check recursion
Zac Medico [Sat, 1 Sep 2012 22:42:45 +0000 (15:42 -0700)]
Eliminate bin/ebuild-helpers/4 directory.
Instead, just use conditionals where appropriate inside
bin/ebuild-helpers/* scripts.
Zac Medico [Sat, 1 Sep 2012 20:51:18 +0000 (13:51 -0700)]
_prepare_self_update: ignore installed version
Now portage will unconditionally make a temporary copy of itself during
updates. This is safer, since it provides maximum protection against
any possible incompatibilities. For example, suppose that the ebuild
has been updated to change the location of installed files (such as
libdir change) without a revision bump.
Zac Medico [Sat, 1 Sep 2012 04:29:24 +0000 (21:29 -0700)]
Enable @module-rebuild for 2.1 branch.
Gregory M. Turner [Fri, 31 Aug 2012 16:34:02 +0000 (09:34 -0700)]
portageq: clarify meaning of uses_root
o s/uses_root/uses_eroot/g -- This name was chosen
before its meaning was changed; update it to reflect its purporse.
o Fix a left-over "<root>" arg-doc that never got updated to "<eroot>"
o If a provided eroot argument was shorter than the ${EPREFIX}, portageq
was truncating it down to nothing and activating relative root,
which means this became equivalent to an explicit argument of
"${PWD}/${EPREFIX}/". This is obviously wrong; and a simple case
of failing to sanity-check inputs from an interface that changed in a
backward-incompatible way.
There's no provision in portageq to query EROOT's that don't end in
${EPREFIX}; so if the user makes such a request, instead of doing
something pathological, dump a helpful error message and bail with
a nonzero exit code.
Signed-off-by: Gregory M. Turner <gmturner007@ameritech.net>
Ulrich Müller [Fri, 31 Aug 2012 15:10:16 +0000 (17:10 +0200)]
EAPI 5: doheader and newheader helper functions
See bug #21310.
Ulrich Müller [Fri, 31 Aug 2012 14:55:16 +0000 (16:55 +0200)]
Update years in header.
Zac Medico [Fri, 31 Aug 2012 14:51:49 +0000 (07:51 -0700)]
Fix usex for EAPI 4-python|4-slot-abi.
Ulrich Müller [Fri, 31 Aug 2012 14:24:09 +0000 (16:24 +0200)]
EAPI 5: usex helper function
See bug #382963.
Zac Medico [Fri, 31 Aug 2012 01:49:10 +0000 (18:49 -0700)]
EAPI 5: --host-root option for best/has_version
See bug #401239 and the PMS patch:
http://git.overlays.gentoo.org/gitweb/?p=proj/pms.git;a=commit;h=
25fb5fca6674215ea8aaa3d0ec3dd3df451eec07
Zac Medico [Thu, 30 Aug 2012 16:43:09 +0000 (09:43 -0700)]
config.setcpv: fix early ret for IUSE_EFFECTIVE
Zac Medico [Thu, 30 Aug 2012 16:33:04 +0000 (09:33 -0700)]
dyn_prepare: init state for apply_user_patches
Zac Medico [Thu, 30 Aug 2012 05:54:51 +0000 (22:54 -0700)]
Remove FEATURES=parse-eapi-ebuild-head
It's already been enabled by default in stable portage for awhile now,
so it should be safe to enable it unconditionally. The PMS eapi-5
branch also says that it's mandatory to parse the EAPI:
http://git.overlays.gentoo.org/gitweb/?p=proj/pms.git;a=commit;h=
91d1e1e39b034bde7e5b981a5616a127135f37fa
Zac Medico [Thu, 30 Aug 2012 05:25:33 +0000 (22:25 -0700)]
EAPI 5: econf --disable-silent-rules
See bug #379497 and the PMS patch:
http://git.overlays.gentoo.org/gitweb/?p=proj/pms.git;a=commit;h=
b7750e67b4772c1064543defb7df6a556f09807b
Zac Medico [Thu, 30 Aug 2012 05:04:42 +0000 (22:04 -0700)]
EAPI 5: User patches (no-op dummy stub for now)
A real apply_user_patches implementation will be a bit more work, so
for now we'll just implement the minimum amount necessary to satisfy
the spec:
http://git.overlays.gentoo.org/gitweb/?p=proj/pms.git;a=commit;h=
a8bf7862967cce36b7f1b408934a774126da2538
Zac Medico [Wed, 29 Aug 2012 20:29:37 +0000 (13:29 -0700)]
EAPI 5: enable slot-operator-deps
Zac Medico [Wed, 29 Aug 2012 20:23:22 +0000 (13:23 -0700)]
Rename slot-abi stuff to refer to slot-operator.
This makes it consistent with the language in the PMS eapi-5 branch:
http://git.overlays.gentoo.org/gitweb/?p=proj/pms.git;a=commit;h=
5d6749ac9e5ddc5b1daaad7737b65fa81c6ece47
Zac Medico [Wed, 29 Aug 2012 16:43:32 +0000 (09:43 -0700)]
EAPI 5: src_test support for parallel tests
See bug #363005 and the PMS patch:
http://git.overlays.gentoo.org/gitweb/?p=proj/pms.git;a=commit;h=
3ec4b3c22582a8ec206bce1e93bab377d7b264b5
Zac Medico [Wed, 29 Aug 2012 16:26:38 +0000 (09:26 -0700)]
EAPI 5: EBUILD_PHASE_FUNC variable
See bug #390765 and the PMS patch:
http://git.overlays.gentoo.org/gitweb/?p=proj/pms.git;a=commit;h=
76ddca560da42fd968c53a2a0c38a6ac840a7ad4
Zac Medico [Wed, 29 Aug 2012 15:51:19 +0000 (08:51 -0700)]
Blacklist IUSE_EFFECTIVE from external env.
Zac Medico [Wed, 29 Aug 2012 07:38:12 +0000 (00:38 -0700)]
EAPI 5: Profile IUSE injection
See bug #176467 and the PMS patch:
http://git.overlays.gentoo.org/gitweb/?p=proj/pms.git;a=commit;h=
d9040ab3482af5f790368bac5d053bf1cd760ba8
Zac Medico [Tue, 28 Aug 2012 03:19:58 +0000 (20:19 -0700)]
Package: skip REQUIRED_USE validation if built
Zac Medico [Mon, 27 Aug 2012 22:16:50 +0000 (15:16 -0700)]
Pass eapi to check_required_use more.
Zac Medico [Mon, 27 Aug 2012 22:13:29 +0000 (15:13 -0700)]
EAPI 5: REQUIRED_USE at-most-one-of ?? operator
See bug #354219 and the PMS patch:
http://git.overlays.gentoo.org/gitweb/?p=proj/pms.git;a=commit;h=
1c2dff2df2305aff88a734e3a2716de1bb69f3b6
Zac Medico [Sun, 26 Aug 2012 22:31:01 +0000 (15:31 -0700)]
_get_eapi_attrs: handle unsupported eapi as None
Zac Medico [Sun, 26 Aug 2012 21:45:05 +0000 (14:45 -0700)]
validate_desktop_entry: filter more kde noise
This will fix bug #432862.
Ulrich Müller [Sat, 25 Aug 2012 21:21:12 +0000 (23:21 +0200)]
Whitespace.
Zac Medico [Sat, 25 Aug 2012 20:40:10 +0000 (13:40 -0700)]
KeywordsManager: fix isStable comment
Ulrich Müller [Sat, 25 Aug 2012 20:00:51 +0000 (13:00 -0700)]
EAPI 5: new* commands can read from stdin, bug 263565.
Zac Medico [Sat, 25 Aug 2012 19:45:21 +0000 (12:45 -0700)]
new*: symlink to generic newins implementation
Zac Medico [Sat, 25 Aug 2012 19:42:28 +0000 (12:42 -0700)]
newins: become generic so new* can be symlinks
Zac Medico [Sat, 25 Aug 2012 04:11:04 +0000 (21:11 -0700)]
EAPI 5 stable use.mask/force for bug #431078
Zac Medico [Sat, 25 Aug 2012 04:08:47 +0000 (21:08 -0700)]
_pkg_str: pass in config + metadata with KEYWORDS
This will be needed in order to support stable use.mask/force for
bug #431078.
Zac Medico [Thu, 23 Aug 2012 22:29:35 +0000 (15:29 -0700)]
prepstrip: avoid debugedit/strip interference
This will fix bug #421099.
Zac Medico [Thu, 23 Aug 2012 17:10:38 +0000 (10:10 -0700)]
config: _eapi_cache.clear() after setcpv
Zac Medico [Thu, 23 Aug 2012 16:55:19 +0000 (09:55 -0700)]
read_corresponding_eapi_file: cache results
Zac Medico [Wed, 22 Aug 2012 21:12:31 +0000 (14:12 -0700)]
Optimize sys.std* flush code.
Zac Medico [Wed, 22 Aug 2012 20:32:28 +0000 (13:32 -0700)]
make.conf.5: s/Hash functions/Hash function/
Zac Medico [Wed, 22 Aug 2012 19:46:04 +0000 (12:46 -0700)]
Use sys.__std*.fileno() in case of overrides.
This fixes AttributeError exceptions for API consumers that override
sys.std* streams pseudo-file objects.
Zac Medico [Wed, 22 Aug 2012 16:23:18 +0000 (09:23 -0700)]
EventLoop: handle EINTR IOError for epoll
Zac Medico [Wed, 22 Aug 2012 05:38:57 +0000 (22:38 -0700)]
EventLoop: use epoll when available
This will fix bug #432024.
Zac Medico [Wed, 22 Aug 2012 04:01:24 +0000 (21:01 -0700)]
_apply_hash_filter: make hash_filter simpler
Now any callable object will work, which might be helpful for
consumers of the Manifest.checkFileHashes() method.
Zac Medico [Tue, 21 Aug 2012 20:22:19 +0000 (13:22 -0700)]
Implement PORTAGE_CHECKSUM_FILTER for bug #432170
Zac Medico [Tue, 21 Aug 2012 03:29:04 +0000 (20:29 -0700)]
PollScheduler: use timeout for loadavg checks
This ensures that the code from commit
fe03b5fd790fc09bd6594c55bb174e80aaac1e5a behaves correctly even when
there are no poll events arriving for a long time (which might happen
if the running builds don't produce output for a long time).
Zac Medico [Mon, 20 Aug 2012 20:37:53 +0000 (13:37 -0700)]
Scheduler: only use poll timeout when needed
This avoids wasteful CPU consumption, as reported in bug #432024.
Dave Kemper [Sun, 19 Aug 2012 21:33:44 +0000 (14:33 -0700)]
make.conf.5: warn about PORTAGE_TMPDIR in PORTDIR
This will fix bug #392565.
Zac Medico [Sun, 19 Aug 2012 05:44:35 +0000 (22:44 -0700)]
emerge-delta-webrsync: tweak "recompressing" msg
FEATURES=webrsync-gpg forces it to recompress in the foreground.
Zac Medico [Sun, 19 Aug 2012 05:21:40 +0000 (22:21 -0700)]
emerge-delta-webrsync: improve date calculations
This borrows some code from emerge-webrsync, and will fix bug #153513.