aboutsummaryrefslogtreecommitdiffstats
path: root/URPM/Resolve.pm
Commit message (Collapse)AuthorAgeFilesLines
* - package promotion must respect strict_archPascal Rigaux2007-09-281-2/+14
|
* prepare next commit:Pascal Rigaux2007-09-281-5/+9
| | | | | | - rename strict_arch_check() into strict_arch_check_installed() - create strict_arch()
* - enhance sorted graph by better taking into account conflicts from ↵Pascal Rigaux2007-09-281-1/+9
| | | | | | | state->{rejected} (fixes "big transaction" in urpmi split-transactions--strict-require.t test)
* - create _add_group() out of sort_graph() for future usePascal Rigaux2007-09-281-25/+26
| | | | | - cleanup
* move things (prepare for next commit)Pascal Rigaux2007-09-281-3/+3
|
* indent debug message so that "conflict above" message can still be meaningfulPascal Rigaux2007-09-281-1/+1
|
* more debug messagePascal Rigaux2007-09-281-3/+11
|
* simplify (don't compute twice $_->fullname ne $p->fullname when we alreadyPascal Rigaux2007-09-281-3/+2
| | | | | know $_->name ne $p->name)
* - handle promotion via obsolete, not only providesPascal Rigaux2007-09-271-0/+12
|
* fix previous commitPascal Rigaux2007-09-271-4/+3
|
* simplifyPascal Rigaux2007-09-271-2/+3
|
* - fix bug doing "urpmi kernel-source"Pascal Rigaux2007-09-201-5/+5
| | | | | ("find" must be seen by perl compiler before using it)
* in case one do "urpmi kernel-devel", $requested will be Pascal Rigaux2007-09-181-1/+1
| | | | | | | { 'id1|id2|...' => 1 }, and not { 'kernel-devel' => 1 } as would be prefered. this need to be changed in urpmi. But we still want to display the correct things here.
* - choose every kernel-<flavor>-devel-<version> packages whenPascal Rigaux2007-09-171-8/+21
| | | | | kernel-<flavor>-<version> is selected
* create _is_selected_or_installed() out of _score_for_locales()Pascal Rigaux2007-09-171-2/+8
|
* simplifyPascal Rigaux2007-09-141-1/+1
|
* - fix regression in 2.00: keep_unrequested_dependencies is still used byPascal Rigaux2007-09-141-1/+6
| | | | | | | installer. restore it, but must now be set trough $urpm->{keep_unrequested_dependencies} - add some more debug info
* - fix bug occurring with --keepPascal Rigaux2007-09-141-8/+5
| | | | | | | (thanks to akurtakov for reporting it on #mandriva-cooker) the fix is to drop field {psel} from {backport}, it's unused anyway
* properly handle the reason for a conflict. Pascal Rigaux2007-09-131-5/+5
| | | | | | this broke in commit "it really seems {why}{conflicts} is only used for debugging purpose, so this change won't break."
* - handle promote from installed package which require a unselected package,Pascal Rigaux2007-09-131-14/+22
| | | | | | | | | | | | whereas new package does not require it anymore (cf test_d & test_e from split-transactions--conflict urpmi test case) the idea is: - resolve_rejected is wrong, since it blindly removes installed pkgs without trying to promote anything - @diff_provides can be nicely used to handle rejecting package one by one - _handle_diff_provides do handle promotion of new packages
* prepare next commitPascal Rigaux2007-09-131-14/+13
|
* rename _handle_provides_overlap() to _handle_conflict()Pascal Rigaux2007-09-131-5/+5
| | | | | (much better name)
* create _compute_diff_provides_of_removed_pkg() out of ↵Pascal Rigaux2007-09-131-1/+8
| | | | _unselect_package_deprecated_by_property()
* fix docPascal Rigaux2007-09-131-1/+1
|
* rename _compute_diff_provides() to _unselect_package_deprecated_by()Pascal Rigaux2007-09-131-7/+7
| | | | | (since it really does much more than computing diff_provides)
* update docPascal Rigaux2007-09-131-2/+2
| | | | | | | | (it changed because of commit: set_rejected() must do things even if _set_rejected_from has been done on the same package. this removes the need for a special case)
* create _add_rejected_backtrack() and use itPascal Rigaux2007-09-131-7/+20
|
* use set_rejected() in _compute_diff_provides_one(),Pascal Rigaux2007-09-121-14/+16
| | | | | nb: the change is not neutral in the "unselected_uninstalled" case
* enhance previous commit: {rejected_pkg} is much better than {required_pkg}Pascal Rigaux2007-09-121-11/+11
|
* resolve_rejected_() and set_rejected() now use a $rdep parameterPascal Rigaux2007-09-121-31/+39
|
* set_rejected() must do things even if _set_rejected_from has been done on thePascal Rigaux2007-09-121-8/+9
| | | | | same package. this removes the need for a special case
* better indicate the test case in urpmiPascal Rigaux2007-09-121-1/+1
|
* update docPascal Rigaux2007-09-121-1/+0
|
* update docPascal Rigaux2007-09-121-2/+3
|
* update docPascal Rigaux2007-09-121-1/+2
|
* create _remove_all_rejected_from() out of disable_selected()Pascal Rigaux2007-09-121-6/+10
|
* create _remove_rejected_from()Pascal Rigaux2007-09-121-4/+16
|
* update docPascal Rigaux2007-09-121-3/+7
|
* factorize code into _set_rejected_from()Pascal Rigaux2007-09-121-7/+13
|
* replacing {backtrack}{closure} with {closure} since it looks like it was a typoPascal Rigaux2007-09-121-2/+2
|
* update docPascal Rigaux2007-09-121-1/+2
|
* - fix typo in previous commit: pass $urpm to set_selected() for debug messagePascal Rigaux2007-09-121-9/+9
| | | | | | - {from} options to resolve_rejected_ and set_selected is now a pkg object instead of a fullname string
* add debug messagePascal Rigaux2007-09-121-5/+6
|
* simplifyPascal Rigaux2007-09-121-8/+4
|
* handle {size} in set_rejected()Pascal Rigaux2007-09-121-8/+11
|
* - add ability to merge {closure} fields in set_rejected (from $options{why})Pascal Rigaux2007-09-121-13/+13
| | | | | | | - {from} is a fullname here - {why} is a hash - use set_rejected() to factorize code
* cleanupPascal Rigaux2007-09-121-2/+2
|
* fix typo in previous commitPascal Rigaux2007-09-121-1/+1
|
* create set_rejected() out of resolve_rejected_()Pascal Rigaux2007-09-121-10/+22
|
* dep field "choices" is here for debug purpose, rename it _choicesPascal Rigaux2007-09-121-1/+1
|
generated by cgit v1.2.1 (git 2.21.0) at 2024-11-24 23:54:34 +0000