40cc5fea1d
Add a few solver tests regarding "replace"
2011-10-15 20:04:29 +02:00
30f6c05069
Add test for choices in circular requirements and delete packagist based test
2011-09-25 23:50:54 +02:00
85f88de59d
Add a solver test for provided virtual packages and circular dependencies
2011-09-25 23:46:24 +02:00
123f5fef77
Refactored Operations in order to fulfill naderman requests
2011-09-25 14:40:12 +03:00
6d667a849e
Remove broken functional test
2011-09-24 17:00:18 +02:00
0a1e7320b0
Refactored solver result array. Now it returns array of operation objects which are much nicer to work with
2011-09-24 02:29:22 +03:00
b1d16a88b7
Test a replace link with a null constraint between packages
2011-08-21 07:08:34 -04:00
69ebbf0eac
Correct handling of rules with more than 2 literals.
2011-08-21 06:30:06 -04:00
ac61c3973b
Add a test for two equivalent installable packages and fix the functional test
2011-08-20 23:05:11 -04:00
0f6fb9b692
Separate transaction generation from solving and properly identify updates
2011-08-20 18:38:31 -04:00
f46aa495da
Turn obsolete relationships and replaced packages into conflict rules
2011-08-20 18:19:47 -04:00
ad25618bdd
Add install job and update rule conflict handling and reenable installation
2011-08-19 06:15:51 -04:00
1125de6bf2
Added disablePolicyRules to solve conflicts between remove jobs & update rules
...
This temporarily breaks installation of packages
2011-08-19 06:06:00 -04:00
01945c20f9
Simplify testSolverWithComposerRepo reusing utility methods from other tests
2011-08-05 04:17:07 -04:00
172c8b908e
Split the large solver test up into individual units testing smaller parts
2011-08-05 04:14:11 -04:00
b277b8e73c
Add failing test with composer repo
2011-07-21 13:42:47 +02:00
825deff09f
Completed analysis of unsolvable situations and main decision process
2011-07-08 07:11:04 -04:00
498125adbe
Assertions are now properly decided before running the main solver.
...
Updating does not work yet. The test case works fine for installing/removing
the correct packages however. Weak (update/feature) rules are entirely ignored
so far.
Watches are on literal ids rather than literals to save on function calls
and memory usage. So a few methods for literals now have an id counter part.
This should probably be refactored to have the literal versions call the
id ones instead.
2011-06-07 22:43:26 +02:00
d181e32000
Fixed test expectation
2011-05-23 02:23:21 +02:00
00260a89ae
Make test fail
2011-05-23 02:18:11 +02:00
7ea2fd6c93
Fix author/copyright
2011-04-17 21:39:53 +02:00
933cc6179b
Import initial partial port of the libzypp satsolver.
2011-04-05 17:37:19 +02:00
Jordi Boggiano