| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
|
|
|
| |
EBUILD_EXIT_STATUS_FILE.
|
| |
|
|
|
|
| |
replacement for EBUILD_EXIT_STATUS_FILE.
|
|
|
|
| |
stdout could corrupt the intended output of this program.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
processes can to communicate with portage's main python process.
Here are a few possible uses:
1) Robust subshell/subprocess die support. This allows the ebuild
environment to reliably die without having to rely on signal IPC.
2) Delegation of portageq calls to the main python process, eliminating
performance and userpriv permission issues.
3) Reliable ebuild termination in cases when the ebuild has accidentally
left orphan processes running in the backgraound (as in bug 278895).
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
| |
in globals() (such as imported modules) as a valid command.
|
|
|
|
| |
Remove the now unneded _src_uri_validate.
|
| |
|
|
|
|
|
| |
This change makes paren_reduce and paren_normalize calls for use_reduce redundant,
resp. wrong. All consumers inside portage are adopted.
|
| |
|
| |
|
| |
|
| |
|
|
|
|
| |
appropriate when necessary.
|
| |
|
| |
|
| |
|
|
|
|
| |
the shebang of configure scripts when appropriate.
|
|
|
|
|
|
| |
where DEPEND is set and RDEPEND is unset in the ebuild, since this
triggers implicit RDEPEND=$DEPEND assignment (prior to EAPI 4) and is
forbidden by the QA team.
|
|
|
|
| |
Thanks to Jeremy Olexa (darkside) for the initial patch.
|
| |
|
| |
|
|
|
|
|
|
|
|
|
| |
for all of the following cases:
* package is missing repository metadata
* package has repository metadata, but the source repository does not
have a profiles/updates/ directory
* package has repository metadata, but the source repository is not
currently accessible via PORTDIR_OVERLAY
|
|
|
|
| |
most efficient to process all repos at once.
|
|
|
|
|
|
| |
As a backwards-compatibility measure, if a particular repository doesn't
come with an 'updates' directory, fallback to updates from
dbapi.porttree_root.
|
| |
|
|
|
|
| |
with python3.1.
|
|
|
|
| |
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
|
|
|
|
| |
It is not used anywhere, and don't seem to be a useful check at all.
|
| |
|
|
|
|
| |
with python2 and python3).
|
|
|
|
|
|
| |
Instead of performing a single 'file.executable' check on auxiliary
files and then another one on ebuild files, use a single check for them.
This unifies the path output too.
|
|
|
|
|
|
|
|
|
|
|
|
| |
environment variables, for use within the chromium-os
build environment.
* Add PORTAGE_USERNAME and PORTAGE_GRPNAME to the environment
whitelist and filter them from the saved environment.
Note: In order to support PORTAGE_USERNAME and PORTAGE_GRPNAME
settings in make.conf, the associated portage.data attributes
may be relocated to the config class in the future.
|
|
|
|
|
| |
Overnight patches have the tendency to be bugged. Now 'repoman --ask'
should work like a normal flag.
|
| |
|
|
|
|
|
|
| |
Put an empty line between the real commit message and Portage version
append. Instead of creating a second pair of parenthesis
for the Manifest commit one, append it to the Portage version.
|
|
|
|
|
|
| |
Instead of using faulty prefix-stripping code on 'git diff-index'
results, simply pass '--relative' to it. Thanks to that, git will do all
the stripping itself and return only changes in the current directory.
|
|
|
|
|
|
|
| |
The '--vcs' option can be used to override VCS autodetection and force
using specific one. It can be especially useful if the work directory is
ambiguous (i.e. there is more than one VCS control directory in the same
directory).
|
|
|
|
|
|
| |
Move the real code into repoman.utilities. Support any repository depth
for distributed SCMs -- i.e. Sunrise through git-svn. Bail out if more
than one control version directory is found at the same depth.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This allows make to do its thing when using Portage in a development
workflow, although the change also exposes how ebuild clean doesn't
clear out the WORKDIR.
TEST=`rm -rf /build/$board/tmp/portage/chromeos-base/kernel-0.0.1 &&
[ -e /build/$board/tmp/portage/chromeos-base/kernel-0.0.1/.unpacked ] &&
rm /build/$board/tmp/portage/chromeos-base/kernel-0.0.1/.unpacked &&
FEATURES=keepwork ebuild-$board kernel-0.0.1.ebuild install`
Then modify some of the source (preferably in the git repository rather
than the WORKDIR).
`[ -e /build/$board/tmp/portage/chromeos-base/kernel-0.0.1/.unpacked ] &&
rm /build/$board/tmp/portage/chromeos-base/kernel-0.0.1/.unpacked &&
FEATURES=keepwork ebuild-$board kernel-0.0.1.ebuild install &&`
On the second ebuild operation, make should be able to tell what
exactly, or if nothing at all, has changed, and rebuild only the
necessary object files.
http://codereview.chromium.org/2922001
|
| |
|
| |
|