summaryrefslogtreecommitdiffstats
path: root/urpm/get_pkgs.pm
Commit message (Collapse)AuthorAgeFilesLines
* revert commit r3849 from Mar 26 2012 as it causes a testsuiteThierry Vignaud2012-06-081-6/+1
| | | | | | | | | regression: "(selected2local_and_blists) prevent having duplicate packages in blists (thus fixing downloading twice noarch packages on x86_64 with --download-all (mga#4867))"
* (selected2local_and_blists) silent perl_checkerThierry Vignaud2012-03-261-1/+1
|
* (selected2local_and_blists) prevent having duplicate packages in blistsThierry Vignaud2012-03-261-1/+6
| | | | | (thus fixing downloading twice noarch packages on x86_64 with --download-all (mga#4867))
* add perl_checker hintsThierry Vignaud2012-02-011-1/+1
|
* perl_checker cleanupsThierry Vignaud2012-01-061-4/+4
|
* Synthesized commit during git-svn import combining previous Mandriva history ↵Mageia SVN-Git Migration2011-09-131-1/+1
| | | | | | | | | | | | with Magiea. This commit consitsts of the following subversion commits: ------------------------------------------------------------------------ r1928 | dmorgan | 2011-09-13 21:44:28 +0100 (Tue, 13 Sep 2011) | 1 line Import urpmi ------------------------------------------------------------------------
* remember which media local packages came fromChristophe Fergeau2010-04-281-1/+6
| | | | | | | | | | | | | | | | | | | | | | It's sometimes possible that a package is available from both a local medium (CDROM) and a distant one. This causes problems when the package isn't signed with the same key (this is bad but happens on MES5 install CDs) as urpmi tends to get confused about where a package comes from and then check the package against the wrong key. Packages are assigned IDs by urpmi, and these IDs are used to figure out the package media. This ID is also used as an identifier to "recognize" identical packages coming from different media, so we need to be careful when manipulating it. In particular, we can't assign the right ID to every package right away because then this would break package downloading (we'd download packages available on local media). When processing packages to download, if we encounter a local package, we assign it the right ID. This assumes the list of media in the blist are sorted by order of relevance (ie prioritary ones first).
* perl_checker fixesChristophe Fergeau2009-10-131-1/+1
|
* check the FS has enough space before trying to download all packagesChristophe Fergeau2009-10-131-0/+21
|
* When downloading packages, remember where it comes from.Christophe Fergeau2009-01-091-4/+11
| | | | | | | | | When the same packages is available from different medias (main/release and main/update for example), urpmi can forget from which media it downloaded the package, and this causes issues with key checking. This commit makes sure we are using the right package id for the package we downloaded (the same package will have different IDs if it's available from different medias).
* Don't modify _download_packages_of_distant_media args, use return valuesChristophe Fergeau2009-01-091-5/+9
|
* add some commentPascal Rigaux2009-01-081-1/+3
|
* add some commentPascal Rigaux2009-01-081-1/+2
|
* use "use strict" in all modulesPascal Rigaux2009-01-061-0/+1
|
* - urpmi, gurpmiPascal Rigaux2008-10-211-1/+3
| | | | | o allow "retry" on aria2 download failure
* - aria2:Pascal Rigaux2008-10-131-0/+1
| | | | | | o do not use --max-file-not-found=3 when downloading rpms (since rpms are "versioned")
* much nicer userdir() functions:Pascal Rigaux2008-09-041-5/+1
| | | | | | | | | - userdir() does not modify $urpm->{cachedir} anymore, if you must ensure $urpm->{cachedir} to be modified, use ensure_valid_cachedir() - valid_cachedir() gives you a valid cachedir (works for root or simple-user), but does not modify $urpm->{cachedir}
* simplify: always create "partial" and "rpms" in "userdir", that way we can usePascal Rigaux2008-09-041-7/+2
| | | | | userdir() in most cases
* - create urpm::blist_to_filenames()Pascal Rigaux2008-09-031-1/+1
| | | | | - use it and sync_rel()
* now, get rid of blist->{list} :)Pascal Rigaux2008-07-081-6/+3
|
* fix typoPascal Rigaux2008-07-081-2/+2
|
* migrate download_packages_of_distant_media() to use blist->{pkgs} instead of ↵Pascal Rigaux2008-07-081-6/+4
| | | | blist->{list}
* migrate _download_packages_of_distant_media() to use blist->{pkgs} instead ↵Pascal Rigaux2008-07-081-7/+8
| | | | of blist->{list}
* add {pkgs} to $blist, it deprecates {list}Pascal Rigaux2008-07-081-3/+7
|
* make selected2list() use selected2local_and_blists()Pascal Rigaux2008-07-081-17/+14
| | | | | and make selected2local_and_blists() do the computation
* rename $i to $id to make it clearPascal Rigaux2008-07-081-6/+6
|
* rename/create vars for clarityPascal Rigaux2008-07-081-5/+6
|
* pass a blist to _download_packages_of_distant_media()Pascal Rigaux2008-07-081-12/+10
|
* re-indentPascal Rigaux2008-07-081-29/+30
|
* create _download_packages_of_distant_media() out of ↵Pascal Rigaux2008-07-081-15/+23
| | | | download_packages_of_distant_media()
* goal: get rid of $list which must be used together with $urpm->{media}, and ↵Pascal Rigaux2008-07-081-8/+27
| | | | | | | | | | | | | | use $blists instead - create urpm::get_pkgs::selected2local_and_blists(), which is similar to selected2list(), but returns a $blists - the following functions have been modified to use a $blists instead of $list: urpm::cdrom::copy_packages_of_removable_media() urpm::install::prepare_transaction() urpm::removable::try_mounting_non_cdroms() urpm::get_pkgs::download_packages_of_distant_media() - adapt code to use $blists
* create cachedir_rpms() out of _selected2local_and_ids()Pascal Rigaux2008-07-081-8/+23
|
* create _selected2local_and_ids() out of selected2list()Pascal Rigaux2008-07-081-9/+20
|
* we don't need to know the medium to compute ->is_delta_installable, so doingPascal Rigaux2008-07-081-4/+4
| | | | | it in previous loop
* rename varPascal Rigaux2008-07-081-3/+3
|
* adapt the error message: --install-src is allowed for non-root for some time,Pascal Rigaux2008-07-071-2/+1
| | | | | | so the real pb here is $cachedir/partial being fucked up (occurs for example when /var is being removed during urpmi...)
* - urpmiPascal Rigaux2008-06-261-1/+1
| | | | | o fix displaying "files are missing" (regression introduced in 5.6)
* empty_dir() is a better name, clean_dir() will be used for just removing the dirPascal Rigaux2008-06-241-2/+2
|
* move removing bad file inside verify_partial_rpm_and_move()Pascal Rigaux2008-02-261-3/+4
| | | | | (it won't hurt urpm::removable::_do_the_copy())
* simplifyPascal Rigaux2008-02-261-2/+2
|
* urpm::get_pkgs::verify_partial_rpm_and_move() return the final rpmPascal Rigaux2008-02-261-1/+1
|
* create verify_partial_rpm_and_move() out ofPascal Rigaux2008-02-261-4/+12
| | | | | download_packages_of_distant_media()
* fix typo (introduced on 2008-02-24 r237346)Pascal Rigaux2008-02-261-1/+1
|
* rename some varsPascal Rigaux2008-02-261-14/+12
|
* display a "bad rpms" message instead of "files are missing" when the ↵Pascal Rigaux2008-02-241-10/+17
| | | | | | | | | | downloaded rpm is bad. this adds a new callback for rpmdrake: "bad_rpms" this modifies the API of urpm::get_pkgs::download_packages_of_distant_media ($error_sources is now { url => reason } instead of { id => url })
* further simplify (allowed by making "moving file" fatal)Pascal Rigaux2008-02-241-2/+1
|
* - urpmi, urpmi.addmedia, urpmi.update:Pascal Rigaux2008-02-241-2/+2
| | | | | | o have fatal errors on some errors that must not happen (eg: moving rpm from download dir to cachedir)
* - urpmi:Pascal Rigaux2008-01-181-3/+11
| | | | | o --install-src as user now works for remote medium
* use var (for future commit)Pascal Rigaux2008-01-181-6/+8
|
* - urpmi:Pascal Rigaux2007-11-151-1/+9
| | | | | | | | o fix "urpmi --install-src" (regression introduced in 4.10.9) nb: packages_by_name can't be used here since $urpm->{provides} doesn't have src.rpm so a full search is needed
generated by cgit v1.2.1 (git 2.21.0) at 2024-11-23 00:12:16 +0000