summaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
* repoman: Check if the prefix.eclass is inherited if eprefixify is used. ↵Sebastian Luther2010-08-044-1/+24
| | | | Thanks to Jeremy Olexa (darkside) for the initial patch.
* Add a sanity check in _preload_portage_submodules() to ensure that theZac Medico2010-08-041-0/+4
| | | | loop always terminates.
* Fix infinite loop in _preload_portage_submodules().Zac Medico2010-08-031-0/+1
|
* Fix typo in docstring.Zac Medico2010-08-031-1/+1
|
* Bug #330645 - Make _unregister_module_proxy() detect partially importedZac Medico2010-08-031-12/+17
| | | | | modules (indicated when an AttributeError is caught) and handle them appropriately.
* Bug #330645 - Use more lazy imports for the portage.data module, inZac Medico2010-08-033-3/+5
| | | | | order to avoid circular import triggered by missing 'portage' user or group.
* Remove code related to $DISTDIR/.locks since that directory isZac Medico2010-08-031-10/+0
| | | | actually unused.
* Make ExtendedAtomDict.__getitem__() return the value directly fromZac Medico2010-08-031-15/+7
| | | | | self._extended when given a wildcard (rather than make an unnecessary copy).
* Move the */* code out of loops for greater efficiency.Zac Medico2010-08-031-19/+19
|
* Make the effect of */* atoms in package.{use,license,properties} show up in ↵Sebastian Luther2010-08-031-1/+19
| | | | emerge --info
* Only populate the binarytree for actions that need it (search orZac Medico2010-08-021-1/+2
| | | | install actions).
* Bug #330937: Handle IOError raised by array.fromfile().Arfrever Frehtes Taifersar Arahesis2010-08-031-1/+2
|
* * Make _world_repo_match() try to match both atoms, so it still worksZac Medico2010-08-011-9/+11
| | | | | if the world file and vdb are slighly out of sync wrt package moves. * Write world file updates at the earliest opportunity.
* Make update_config_files() take a dict of {repo_name: list} since it's moreZac Medico2010-08-012-24/+54
| | | | | | efficient this way, and this function can't be called separately for each repo if /etc/portage is under CONFIG_PROTECT (because it would produce separate updated files for each repo).
* Fix NameError for ConfigProtect.Zac Medico2010-08-011-1/+1
|
* Just return boolean to indicate whether or not updates have been performed,Zac Medico2010-07-311-10/+8
| | | | since that's all that the calling code currently needs.
* Move the mtimedb timestamps code out of an unnecessary loop.Zac Medico2010-07-311-2/+3
|
* Always use shlex_split() to split CONFIG_PROTECT{,_MASK}.Zac Medico2010-07-315-19/+25
|
* Add a warning message to the --package-moves[=n] docs, noting thatZac Medico2010-07-282-2/+8
| | | | should remain enabled under normal circumstances.
* Add a --package-moves[=n] option that can be used to control theZac Medico2010-07-284-7/+29
| | | | | "Performing Global Updates" routine. This option is enabled by default.
* Document --depclean-lib-check[=n].Zac Medico2010-07-282-0/+14
|
* Fix calc_depclean() to skip the topological sort code when there'sZac Medico2010-07-281-1/+1
| | | | nothing selected for removal.
* Tweak global updates handling so that updates from $PORTDIR are appliedZac Medico2010-07-276-33/+102
| | | | | | | | | 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
* Fix UnboundLocalError for do_upgrade_packagesmessage.Zac Medico2010-07-271-1/+1
|
* Fix UnboundLocalError for timestamps variable.Zac Medico2010-07-271-1/+1
|
* Make dbapi.update_ents() take a dict of {repo_name: list}, since it'sZac Medico2010-07-273-16/+31
| | | | most efficient to process all repos at once.
* Fallback to updates from PORTDIR if repository has no 'updates' dir.Michał Górny2010-07-273-0/+9
| | | | | | As a backwards-compatibility measure, if a particular repository doesn't come with an 'updates' directory, fallback to updates from dbapi.porttree_root.
* Add repository-wide updates support to emaint.Michał Górny2010-07-271-40/+55
|
* Support repository-wide updates in FakeVartree.Michał Górny2010-07-271-13/+24
|
* Limit repository-wide dbentry updates.Michał Górny2010-07-272-3/+7
| | | | | Update dbentries only when the repository the package comes from matches the repository update comes from.
* Support repository-wide updates.Michał Górny2010-07-275-126/+165
| | | | | | | Support reading repository-wide update files. Apply them as long as the 'repository' for vardb entry matches the repository update is originating from. For world and configuration files, use the best vardb match.
* Make the world-candidate package set attribute default to False, sinceZac Medico2010-07-274-18/+6
| | | | [usersets] is the only one for which it is True in the default config.
* extended atom syntax: Add comment to make.conf man pageSebastian Luther2010-07-271-2/+7
|
* Tests: add sets/base/testInternalPackageSetSebastian Luther2010-07-273-0/+61
|
* portage.sets: Don't allow extended atoms unless explicitly told toSebastian Luther2010-07-272-5/+13
|
* Use a directory for the default set configuration.Michał Górny2010-07-275-11/+20
| | | | | | | Expect /usr/share/portage/config/sets to be a directory containing any number of set configuration files. The default Portage sets.conf should be now installed as sets/portage.conf, and other ebuilds are free to install their own set configuration files there.
* Fix PackageSet._updateAtomMap() to use ExtendedAtomDict.setdefault()Zac Medico2010-07-271-2/+1
| | | | | since ExtendedAtomDict__getitem__ doesn't allow modification of items.
* Fix race condition in Portage symlink creation.David James2010-07-261-4/+12
|
* Fix ExtendedAtomDict so get() and __getitem__() behave consistently,Zac Medico2010-07-262-14/+28
| | | | since otherwise it can be quite confusing.
* Make ExtendedAtomDict inherit from MutableMapping, and fix __getitem__Zac Medico2010-07-261-1/+6
| | | | to raise KeyError if given a non-string argument.
* Allow the --depclean library consumer check to be disabled byZac Medico2010-07-262-2/+13
| | | | --depclean-lib-check=n.
* If more than 20 file collisions occur, only look up the first 20 sinceZac Medico2010-07-261-0/+4
| | | | get_owners is slow for large numbers of files.
* When the iter_owners dblink cache becomes full, do not finish processingZac Medico2010-07-261-36/+35
| | | | the current path, and go directly to the low-memory implemention.
* Make extended_cp_match() use re.escape() for safety, and since the result isZac Medico2010-07-251-2/+2
| | | | cached here anyway.
* Bug #329159 - If there are merge collisions are detected, only search forZac Medico2010-07-251-21/+23
| | | | owners if either collision-protect or protect-owned is enabled.
* Make iter_owners() switch to a slower but lower-memory mode if the numberZac Medico2010-07-251-2/+55
| | | | | of given files > 10 or dynamically if the dblink_cache grows to 20 instances.
* Bug #329803 - Make unmerge() show a machine-readable list of selected packages.Sebastian Luther2010-07-251-0/+2
|
* Make extended_cp_match() cache regular expressions for re-use.Zac Medico2010-07-251-2/+8
|
* Add a gc.collect() call inside iter_owers(), since people are reportingZac Medico2010-07-251-0/+2
| | | | | | high memory usage there. We should really only use this function to search for a small number of files. Larger numbers of files should use a different algorithm that will ensure that each CONTENTS file is only parsed once.
* Re-implement extended_cp_match() using regex, and only support cp (insteadZac Medico2010-07-251-42/+8
| | | | of full atoms) since that all we need.
generated by cgit v1.2.3-1-g7c22 (git 2.25.1) at 2024-07-04 09:42:16 +0000