| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This patch series adds support for using a portage installed in one prefix
to build packages with a different prefix.
The current portage has a single EPREFIX variable specifying both the prefix
of the portage installation, and the prefix of the packages portage is
building. This patch series splits it into two parts: the
portage.const.EPREFIX variable specifying the prefix of the portage
installation, used for constructing the PATH as well as the paths to files
belonging to a portage installation itself rather than a target root; and the
EPREFIX setting in config instances, specifying the prefix of the
to-be-built packages and being used for almost everything else.
The EPREFIX config setting defaults to const.EPREFIX, but can be overridden
by the EPREFIX environment variable, as well as the emerge --prefix option.
This allows one to install systems with different prefixes using
`EPREFIX=/foo emerge @system`, though some unrelated changes need to happen
elsewhere first in order to make that a reality.
Ruud Koolen (3):
Distinguish between portage prefix and package prefix
Based GLOBAL_CONFIG_PATH and DEPCACHE_PATH on portage prefix
Pick up EPREFIX environment variable
|
|
|
|
|
| |
This ensures that the main-repo default priority is set to -1000
in all cases where it's appropriate.
|
| |
|
| |
|
| |
|
|
|
|
|
|
| |
This autoclean invocation would should have no effect anyway, unless
the vdb happens to be in an inconsistent state due to a previously
interrupted emerge (relatively rare).
|
|
|
|
|
|
|
|
| |
Show parents for packages involved in blocker conflicts but not
directly involved in slot conflicts, since this information may be
helpful for troubleshooting blocker conflicts. This information had
been hidded when there were unrelated simultaneous slot conflicts,
since commit 77b651300731ec007cd535a83b8ee9a898602783.
|
| |
|
| |
|
|
|
|
|
| |
Instead of suggesting to call dispatch-conf, call chk_updated_cfg_files
in order to display a config update message when appropriate.
|
| |
|
| |
|
|
|
|
|
| |
There's no need to copy the list before calling dep_wordreduce, since
dep_wordreduce copies the list internally.
|
| |
|
| |
|
|
|
|
|
|
|
|
|
| |
This allows emerge_main to avoid having direct local references to
large local objects (like "settings" and "trees"), making it easier to
ensure that stale objects can be garbage collected when other functions
refresh the config with calls to load_emerge_config(). This will be
much more flexible than the "gc_locals" approach that was introduce in
commit e9fd283aedf54e2effc73f4157524fe9a26993c0.
|
| |
|
|
|
|
|
|
|
| |
The python interpreter does _not_ guarantee that destructors are
called for objects that remain when the interpreter exits, so we
use an atexit hook to call destructors for any global portdbapi
instances that may have been constructed.
|
| |
|
|
|
|
| |
Now _match_use does not have to construct them itself.
|
|
|
|
|
|
|
|
| |
This allows portdbapi.portdbapi_instances to be eliminated, which is
nice because we no longer has to be so careful to avoid memory leaks
involving this variable. It was not just annoying for portage
internals, but also for any API consumers that needed to create/destroy
many portdbapi instances.
|
|
|
|
|
|
|
|
|
|
| |
_emerge.Package.Package._iuse.get_real_flag().
It fixes support for USE aliases in:
best_version
has_version
emerge --keep-going
emerge --resume
|
|
|
|
| |
pkg argument of type str.
|
|
|
|
|
|
| |
The CHOST of a binary package does not necessarily match the make.conf
setting, especially if ACCEPT_CHOSTS is configured to match other CHOST
values.
|
|
|
|
|
|
|
| |
This factors --tree logic out of the calling code, and allows
optimization of _show_merge_list to use reference comparison instead
of == comparison. Also, deprecate the unused depgraph.altlist()
"reversed" parameter, due to builtin name collision.
|
|
|
|
|
| |
TODO: Optimize _show_merge_list to use reference comparison more
instead of == comparison.
|
| |
|
| |
|
|
|
|
| |
This shares/aligns logic with is_valid_package_atom.
|
|
|
|
|
|
|
| |
Before, a wildcard atom could be treated as valid even though it
specified a version without an operator, as reported at here:
https://forums.gentoo.org/viewtopic-p-7314746.html#7314746
|
|
|
|
|
|
| |
This adds support for repo.eapi.banned (fatal) and repo.eapi.deprecated
(warning) checks which are controlled by eapis-banned and
eapis-deprecated settings in a repository's metadata/layout.conf.
|
|
|
|
|
|
| |
Before, the file name would be passed directly to FETCHCOMMAND as
though it were a valid URI. Now, FETCHCOMMAND will only be called when
there is a valid URI or a mirror to try.
|
| |
|
| |
|
|
|
|
|
|
|
| |
Additional ssh options to be used when portage executes ssh or sftp.
This variable supports use of embedded quote characters to quote
whitespace or special shell characters within arguments (embedded
quotes must be escaped in make.conf settings).
|
|
|
|
| |
This will fix bug #470258.
|
| |
|
| |
|
|
|
|
|
| |
This removes obsolete parallel-fetch code which is no longer needed,
since EbuildFetcher no longer calls doebuild.
|
|
|
|
|
| |
Do not generate the sftp command for downloading the Packages file, and
simply use the FETCHCOMMAND_SFTP setting.
|
|
|
|
|
| |
Don't use urlopen for https, since it doesn't support
certificate/hostname verification (bug #469888).
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Newer prelinks can support /etc/prelink.conf.d/ files. So that prelink
can install /etc/prelink.conf and manage it itself, have env-update only
write /etc/prelink.conf.d/portage.conf instead of clobbering the main
/etc/prelink.conf file.
This should be backwards compatible as portage will conditionally change
/etc/prelink.conf to use the new /etc/prelink.conf.d/ too.
URL: http://bugs.gentoo.org/266855
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
|
| |
|
| |
|
| |
|
|
|
|
| |
PROPERTIES is not allowed to contain ||, according to PMS.
|
| |
|