| Commit message (Collapse) | Author | Age | Files | Lines |
... | |
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
| |
The EPREFIX variable is not readonly in EAPI 0, 1, and 2, so we can
declare EPREFIX as a local variable. We must be careful not to let this
variable leak into the environment of ebuild/eclass functions that we
call, since bash propagates local variables to nested function calls.
|
|
|
|
|
|
|
| |
The ED variable is not readonly in EAPI 0, 1, and 2, so we can declare
ED as a local variable. We must be careful not to let this variable
leak into the environment of ebuild/eclass functions that we call,
since bash propagates local variables to nested function calls.
|
| |
|
|
|
|
|
| |
This relocates the PATH generation code from ebuild.sh to
doebuild_environment, which helps to eliminate duplicate code.
|
| |
|
|
|
|
| |
Hopefully this helps us avoid errors in loop logic like bug #389047.
|
| |
|
|
|
|
|
| |
In the context of ebuilds, ROOT=/ override is the only common case, so
it makes sense to only support ROOT overrides.
|
|
|
|
|
|
| |
This has been broken since commit
ab484dc9a2612aa6709fad3ff926c8589a706637. In order to support prefix,
callers will have to override EROOT instead.
|
| |
|
|
|
|
|
|
|
| |
It should be safe to assume that REQUIRED_USE is satisfied if the
package is built, so it's a waste to save it. Also, fix code which
assumes that built Package instances have a REQUIRED_USE key in their
metadata.
|
| |
|
|
|
|
|
|
|
| |
dyn_pretend
helpers_die
nonfatal
register_success_hook
|
|
|
|
|
|
| |
This fixes a case where empty PORTAGE_COMPRESS made ecompressdir forget
to call restore_skip_dirs, such that docompress -x resulted in files
not being installed due to hide_skip_dirs hiding them permanently.
|
| |
|
| |
|
|
|
|
| |
This allows our prefix tests to use helpers like insinto, doins, and dosym.
|
| |
|
| |
|
|
|
|
|
| |
We can use the raise_stat_collision attribute to determine when it is
necessary to check for redundant writes.
|
| |
|
|
|
|
|
| |
The pms cache already does this automatically, since __setitem__ calls
are used to detect stat collisions in order to solve bug #139134.
|
|
|
|
|
|
|
| |
The GenCache class assumes that portdb.porttrees[0] is the desired
tree, so assert that it is true. The behavior should have already been
correct, since PORTDIR_OVERLAY was already set to an empty string when
necessary.
|
|
|
|
|
| |
This has been broken since the _eclasses_ structure changed in commit
2ed1cb53cc4158af08c22d466b15b9a9a7767212.
|
| |
|
|
|
|
|
| |
If layout.conf contains "cache-formats = md5-dict pms", egencache will
now write both formats.
|
|
|
|
|
|
| |
Since commit a715b65f7bd36409c1283e6911265d1f4405ab7a, portageq takes
EROOT as the first argument of commands that require a root argument.
So, update the docs to refer to <eroot> instead of <root>.
|
| |
|
|
|
|
|
| |
This is preferred since the bytes type is available in all supported
python versions, while the unicode type is only available in python2.
|
|
|
|
|
|
| |
This simplifies the checks since the bytes type is available in all
supported python versions, while the unicode type is only available in
python2.
|
|
|
|
| |
These go along with commit a715b65f7bd36409c1283e6911265d1f4405ab7a.
|
| |
|
|
|
|
|
|
| |
We read layout.conf cache-formats from left to right and use the first
supported type that's found. This will allow support for multiple
formats in parallel, providing for smooth transitions between formats.
|
|
|
|
|
|
|
|
|
|
| |
It makes more sense to use EROOT instead of ROOT for keys in mappings
like portage.db, since it allows for multiple prefixes to exist
simultaneously within the same map without having a key collision.
This affects all portageq commands which take a <root> parameter, since
that parameter now corresponds to EROOT instead of ROOT. None of this
makes any difference for non-prefix users, since EROOT and ROOT are
identical when EPREFIX is empty.
|
| |
|
| |
|
|
|
|
|
|
|
| |
Use skel.ChangeLog from the repo to create the header of a new ChangeLog
file. Else, we just retain the original header of the ChangeLog. When
no skel.ChangeLog file exists, and this is a new ChangeLog, no header is
used.
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Updating the ChangeLog file may be desirable in more repos than just the
one named 'gentoo', like e.g. the Prefix one. Hence, make this default
configurable though metadata/layout.conf.
This commit must go accompanied by a commit to
gentoo-x86/metadata/layout.conf that adds the following bit:
# Bug #337853 - gentoo's council says to enable
# --echangelog by default for the "gentoo" repo
update-changelog = true
|
|
|
|
|
|
| |
Having sleep in the world-writable loop causes riduculous delays for
prefix users with exotic filesystem constraints (like a FAT filesystem
mounted with umask=000).
|
| |
|
|
|
|
|
| |
Allows forced ChangeLog generation even when the vcs has detected that
the ChangeLog has already been modified.
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
| |
Instead of calling echangelog, which on its turn has to query the VCS
again, use the existing information on changes made to the current
directory, and update the ChangeLog from Python itself.
This avoids a call to echangelog, and avoids again retrieving the same
VCS information as repoman already did. It makes repoman independent
from external tools it didn't install itself, and should be faster in
general.
|
| |
|