summaryrefslogtreecommitdiffstats
path: root/urpm/select.pm
Commit message (Collapse)AuthorAgeFilesLines
* - urpmq:Pascal Rigaux2008-01-181-2/+6
| | | | | | | | | | | | o fix option -a : display packages of all compatible archs (#36942) this fixes another issue introduced in commit r231535 (urpmi 4.10.15, 2007-11-15), partially fixed in r231782 (2007-11-28) the commit has 2 parts: - ensure all compatible archs are kept when {all} option is used - display one pkg per line in urpmq
* better fix for "always prompt before doing a priority-upgrade transaction,Pascal Rigaux2008-01-111-1/+1
| | | | | | | | | | | | even if there is only one priority upgrade package (since there will be more packages to install after restarting urpmi)" since we don't want to prompt if all pkgs have been especially prompted on command-line also not using suggests when selecting pkgs in priority-upgrade (nb: this may by-pass some suggests in priority-upgrade pkgs, need more checking)
* (find_packages_to_remove) fix commentThierry Vignaud2007-12-061-1/+1
|
* - urpmi:Pascal Rigaux2007-11-301-1/+1
| | | | | | | o fix regression introduced in 4.10.15: "urpmi --force valid invalid" should warn about "invalid" but still install "valid"
* ensure "urpmq --fuzzy" displays each packages alone on its line.Pascal Rigaux2007-11-281-2/+6
| | | | | | | | | | | | | | this regression was introduced in commit r231535: there is no reason search_packages should return multiple pkgs with same fullname when using {provides}, and should not otherwise a "better" fix could be done in urpmq. nb: "urpmq -d basesystem" still use "|" for choices, but "urpmq --sources -d basesystem" doesn't, so it's quite broken...
* - urpmq:Pascal Rigaux2007-11-281-4/+3
| | | | | o bugfix previous release: "urpmq --fuzzy foo" should display all "*foo*" matches
* remove unneeded codePascal Rigaux2007-11-151-9/+0
| | | | | (this fixes pubkey checks when a src.rpm is available in more than one media)
* fix resolve_dependencies for install_src Pascal Rigaux2007-11-151-1/+1
| | | | | | | (broken since search_packages can return more than one src.rpm) anyway, this code is not needed and will be dropped
* factorize some code by creating search_packagesPascal Rigaux2007-11-151-16/+26
|
* there is no reason search_packages should return multiple pkgs with samePascal Rigaux2007-11-151-3/+6
| | | | | | | fullname when using {provides}, and should not otherwise (will cleanup duplication in next commit)
* - rename $_ into $vPascal Rigaux2007-10-151-8/+11
| | | | | - add some debug (it may help dvalin)
* use existing function (thanks titi)Pascal Rigaux2007-09-271-2/+2
|
* reverting commit (13 March 2006): Pascal Rigaux2007-09-131-6/+5
| | | | | | | "Passing bundle option, even if that does nothing for now" since it does nothing. will revert if/when support is really implemented
* simplifyPascal Rigaux2007-09-111-3/+1
|
* unrequested doesn't exist, removing itPascal Rigaux2007-09-111-2/+0
|
* fix error in previous commitPascal Rigaux2007-09-101-1/+1
|
* if there is a prefered package, it's the first onePascal Rigaux2007-09-101-1/+5
|
* fix typo in previous previous commitPascal Rigaux2007-09-061-1/+1
|
* handle --searchmedia <media1>,...,<mediaN>Pascal Rigaux2007-09-061-3/+9
| | | | | (ie more than one searchmedia)
* new functions:Pascal Rigaux2007-09-061-8/+19
| | | | | | - use URPM::build_listid - prepare for next commit which will handle more than one searchmedia
* - urpmiPascal Rigaux2007-08-281-0/+19
| | | | | | | o handle changes in priority-upgrade list between old and new urpmi (#32925) nb: keep --no-priority-upgrade since it is passed by older urpmi when restarting
* (translate_why_removed_one) if we got back an explanation unknown ofThierry Vignaud2007-08-131-0/+2
| | | | | use, just report it back
* (translate_why_unselected_one) try harder to explain why we cannotThierry Vignaud2007-08-131-1/+4
| | | | | select a package (eg: because i586 package is already installed on x86_64)
* - libraryPascal Rigaux2007-08-121-2/+2
| | | | | | | o urpm::select: new function get_preferred() replacing sort_choices() (for drakx) o get rid of "use of unitialized" when file is unset (for drakx)
* - urpmiPascal Rigaux2007-08-111-0/+33
| | | | | | o handle preferred choices (through --prefer, /etc/urpmi/prefer.list and /etc/urpmi/prefer.vendor.list)
* - urpmiPascal Rigaux2007-08-091-0/+1
| | | | | | | | | | | o explicit error when using "urpmi ---install-src" as user with remote media o add support for "suggests": a newly suggested package is installed as if required, but one can remove it afterwards, or use --no-suggests - urpmf o handle --suggests all the work is done in perl-URPM 1.69
* fix previous "prefer best architecture over exact name" commit:Pascal Rigaux2007-06-081-2/+3
| | | | | | | noarch pkgs need to be taken into account differently (nb: is it a pb that the list order is not kept?)
* - urpmiPascal Rigaux2007-06-071-0/+10
| | | | | | | o prefer best architecture over exact name (eg: urpmi libfoo-devel prefers lib64foo-devel over libfoo-devel) - add function min()
* re-sync after the big svn lossPascal Rigaux2007-04-241-29/+40
|
* log rpmdb search for removalPascal Rigaux2006-11-301-0/+2
|
* directly set $urpm->{root}, it's simplerPascal Rigaux2006-11-291-2/+1
| | | | | (also use it in urpm::get_pkgs::selected2list() even if callers didn't pass it)
* move uniq to urpm::util (i need it in urpm::media)Pascal Rigaux2006-11-281-2/+0
|
* move many functions from urpm.pm to new module urpm/select.pmPascal Rigaux2006-11-211-0/+442