summaryrefslogtreecommitdiff
path: root/src
AgeCommit message (Collapse)AuthorFilesLines
2013-06-04add POOL_FLAG_IMPLICITOBSOLETEUSESCOLORS to match the current rpm behaviourMichael Schroeder6-7/+15
Sigh. I hope that's the last time rpm changes the way it handles obsoletes.
2013-06-03add POOL_FLAG_ADDFILEPROVIDESFILTERED option to make pool_addfileprovides ↵Michael Schroeder4-18/+180
only add files in the standard locations Non-standard files are searched lazyly on demand.
2013-05-29repo_rpmmd: Parse optional baseurl.Zdenek Pavlas1-0/+1
2013-05-28datamatcher_init: play it safe and do a strdup on the match stringMichael Schroeder1-1/+3
2013-05-27support a special Datapos that describes the repodata meta sectionMichael Schroeder1-5/+9
2013-05-27add pool_lookup_idarray for completenessMichael Schroeder4-0/+26
2013-05-17allow SOLVID_META queries through SOLVID_POSMichael Schroeder2-18/+27
Add 'meta' attribute in repo bindings and get rid of no longer neededed lookup methods.
2013-05-15Lookat at keepexplicitobsoletes when creating update rules for multiversion ↵Michael Schroeder1-1/+8
packages
2013-05-15fix multiversion handling when the update rule is an assertionMichael Schroeder1-11/+11
2013-05-07add SOLVER_RULE_JOB_UNSUPPORTED and SOLVER_RULE_JOB_UNKNOWN_PACKAGEMichael Schroeder4-6/+16
2013-05-06add utf8 helpers to util.cMichael Schroeder3-0/+124
2013-04-26bring libsolv in line with currrent rpm's multiversion handlingMichael Schroeder4-2/+23
A multiversion install still obeys the obsoletes of a package. The solver already had a flag to support this, "keepexplicitobsoletes", but as this is a feature of the package manager it belongs in the pool, so that the transaction code can do the right thing. We now have a "noobsoletesmultiversion" flag in the pool that also defines the default for "keepexplicitobsoletes".
2013-04-24solverdebug: add solver_get_orphaned call that somehow got lostMichael Schroeder1-0/+1
2013-04-24terminate solver_create_decisions_obsoletesmap functionMichael Schroeder4-99/+0
2013-04-24move stringification functions from solverdebug to more correct placesMichael Schroeder10-472/+358
2013-04-24use SOLVER_TRANSACTION_OBSOLETE_IS_UPGRADE in transaction_printMichael Schroeder1-1/+1
2013-04-24add SOLVER_TRANSACTION_OBSOLETE_IS_UPGRADE helper flagMichael Schroeder2-2/+8
Many users don't want to show the obsoletes as extra part of the classification. Only makes sense when SOLVER_TRANSACTION_SHOW_OBSOLETES is also specified.
2013-04-23add repodata_translate_id to get rid of some "implement me" assertionsMichael Schroeder3-0/+20
2013-04-23add repo_add_rpmdb_reffp to easily add rpmdb content when there is an old ↵Michael Schroeder3-0/+50
solv file
2013-04-22add pool_setnamespacecallback() and pool_flush_namespaceproviders()Michael Schroeder3-0/+32
2013-04-22small cleanup in id2strMichael Schroeder1-4/+2
2013-04-18selection_filter(): Don't leak Queue q1 and Map m2Ingo Weinhold1-1/+7
2013-04-18Haiku: The notation for the pre-release version part changedIngo Weinhold3-20/+2
Instead of using '-' as the separator for the pre-release '~' is used now. This allows to drop code in a few places where Haiku had to be special-cased before.
2013-04-17rules/selection: use strrchr in EVR checks as wellMichael Schroeder2-2/+2
2013-04-17split arch from the right for debian/haiku (thanks Ingo!)Michael Schroeder1-2/+2
2013-04-17adapt SET_EV/SET_EVR magic to haiku for completenessMichael Schroeder2-6/+10
2013-04-17make sure releases start numeric for haikuMichael Schroeder1-1/+14
otherwise a pre_release might be interpreted as a revision
2013-04-16make pool_solvable2str use -<arch> for haikuMichael Schroeder1-1/+1
2013-04-16support haiku package naming in selection_canonMichael Schroeder1-0/+29
2013-04-16switch range->name and range->evr in REL_COMPAT deps so that stringification ↵Michael Schroeder1-6/+6
works
2013-04-16add DISTTYPE_HAIKU, some small fixesMichael Schroeder5-19/+52
- ARCH_NOARCH -> ARCH_ANY - use - as prerelease separator - correct usage of solv_vercmp - stringify REL_COMPAT, use != instead of <>
2013-04-16Add support for Haiku compat version comparisonIngo Weinhold2-0/+14
2013-04-16Add Haiku semantics version comparisonIngo Weinhold1-0/+96
2013-04-08fix bugs in fileconflicts code and prepare for aliased dirsMichael Schroeder1-2/+2
Breaks API but that should not be a problem as noone seems to use the function yet.
2013-04-05cleanup hash code, it makes no sense to have an extra type for the maskMichael Schroeder7-16/+8
2013-04-05make queue_insertn more flexibleMichael Schroeder6-13/+11
Yes, it's an API change, but I'm pretty sure that nobody other than the solver itself uses queue_insertn.
2013-04-05refactor a bitMichael Schroeder1-6/+17
2013-04-04Do extra checking for choice rules so that they don't interfere with package ↵Michael Schroeder1-5/+85
splits
2013-04-02Make all public libsolv headers C++ safeIngo Weinhold23-17/+165
A few headers already included 'export "C"' blocks, also wrapping other #include's. This patch cleanly wraps the contents of all public headers instead.
2013-03-30solv_sort(): Fix build for non-glibc without qsort_r()Ingo Weinhold1-8/+6
2013-03-30Improve C89 complianceIngo Weinhold1-1/+2
Make sure variables are declared at the beginning of a block.
2013-03-27Support aarch64Dirk Mueller1-0/+1
2013-03-26Fix misuses of memsetAndreas Schwab1-1/+1
src/md5.c: In function 'sat_MD5_Final': src/md5.c:269:23: error: argument to 'sizeof' in 'memset' call is the same expression as the destination; did you mean to dereference it? [-Werror=sizeof-pointer-memaccess] memset(ctx, 0, sizeof(ctx)); ^ examples/solv.c: In function 'read_repos': examples/solv.c:1616:27: error: argument to 'sizeof' in 'memset' call is the same expression as the destination; did you mean to remove the addressof? [-Werror=sizeof-pointer-memaccess] memset(&stb, 0, sizeof(&stb)); ^
2013-03-22rename "NOOBSOLETES" to "MULTIVERSION"Michael Schroeder11-88/+99
The name was always not very fitting, but newer rpm versions do look at the obsoletes and just don't remove packages with the same name, which makes the old name even worse. So rename to "multiversion", which seems to be what many package managers use anyway.
2013-03-22turn splitprovides off when doing the alreadyrecommended checkMichael Schroeder1-1/+4
2013-03-20improve pool_solvable2str output if the evr or arch is emptyMichael Schroeder1-4/+10
2013-03-06cleanup code now that the vendorcheck callback is in the poolMichael Schroeder3-23/+29
2013-03-06get rid of the ugly policy callbacks while we're breaking the ABI, add new ↵Michael Schroeder5-69/+14
pool_set_custom_vendorcheck function
2013-03-06shuffle some ids now that we break the ABIMichael Schroeder1-15/+13
2013-03-06fix dataiterator returning random data in some casesMichael Schroeder5-7/+136
Fixed two issues: 1) iterating over attributes in with vertical storage could return junk when some other operation paged in some other data. 2) the returned string could also be in the paged area, or also in tmpspace if stringification was done. We can't simply alloc memory as this makes things slower, but in most cases the storage does not matter. So we offer a new function, dataiterator_strdup(), that makes the kv.str pointer persistent. WARNING: this commit is an ABI change as it changes the dataiterator structure. Sorry.