summaryrefslogtreecommitdiffstats
path: root/urpm/media.pm
Commit message (Collapse)AuthorAgeFilesLines
* - factorize code from _update_medium__parse_if_unmodified__remote() to make itPascal Rigaux2006-11-281-28/+53
| | | | | | similar to _update_medium__parse_if_unmodified__local() - rationalize error handling
* factorizePascal Rigaux2006-11-281-4/+2
|
* we don't need to pass $hdlist_or to ↵Pascal Rigaux2006-11-281-5/+5
| | | | _read_existing_synthesis_and_hdlist_if_same_time_and_msize() anymore :)
* stop using {hdlist} name in cache/partial, use the downloaded namePascal Rigaux2006-11-281-23/+20
| | | | | | (otherwise things are handled differently for local and remote, and the partial file name for synthesis.hdlist.cz is hdlist.cz which is misleading)
* - create cachedir_with_hdlist() and use it to get rid of $basenamePascal Rigaux2006-11-271-24/+22
| | | | | | - _read_existing_synthesis_and_hdlist_if_same_time_and_msize() now takes a hdlist_or instead of the basename
* fix previous commit (it was the case where {hdlist} was unset and we set it)Pascal Rigaux2006-11-271-14/+11
|
* when parsing virtual media, do check for new missing provides, those may comePascal Rigaux2006-11-271-5/+7
| | | | | from non virtual media
* fix display error message when need_hdlist and no hdlistPascal Rigaux2006-11-241-2/+5
|
* parse a statedir synthesis if it exists, even on virtualPascal Rigaux2006-11-241-14/+19
| | | | | (preparing for having a synthesis built on virtual & hdlist)
* use _parse_hdlist_or_synthesis()Pascal Rigaux2006-11-241-13/+7
|
* create _parse_hdlist_or_synthesis() and use it.Pascal Rigaux2006-11-241-10/+16
| | | | | | (nb: {must_build_synthesis} can still be set on {virtual} media, it won't do anything... for now :)
* do display the errorPascal Rigaux2006-11-241-2/+3
|
* remove defensive programming checks, which are already done inPascal Rigaux2006-11-241-17/+6
| | | | | check_existing_medium() or add_medium()
* remove check already done in check_existing_medium()Pascal Rigaux2006-11-241-36/+39
|
* better var name ($medium is always used to hold a medium, not a number)Pascal Rigaux2006-11-241-3/+3
|
* cleanerPascal Rigaux2006-11-241-0/+1
|
* rename var $second_pass into $need_second_pass (more clear)Pascal Rigaux2006-11-241-16/+17
|
* - move lock functions to a full blown modulePascal Rigaux2006-11-231-11/+5
| | | | | | | | | | | | - handle get_exclusive and release_exclusive on existing lock, *but* don't use it for now! (not useful) - remove urpmi_db locking from add_medium and update_media - move rpm_db locking into _read_cachedir_pubkey - update_media: restrict rpm_db exclusive locking to parse_pubkeys - urpmi: move locking rpm_db near its use - handle lock directly in urpmi.addmedia, urpmi.update, urpme, urpmf - log locking/unlocking
* don't have a global variable to hold locks, otherwise code can't lock twicePascal Rigaux2006-11-231-6/+7
| | | | | non-exclusive
* remove md5sum from @PER_MEDIA_OPT (which only list per-medium options inPascal Rigaux2006-11-221-2/+2
| | | | | | | | | | urpmi.cfg), so that urpmi.cfg do not contain the md5sum (and so urpmi.cfg doesn't need to be modified on urpmi.update, only /var/lib/urpmi/MD5SUM is modified) (bug introduced when creating @PER_MEDIA_OPT in r35745, fixed in r35832, but broken again in r35953)
* create urpm::removable out of urpm.pmPascal Rigaux2006-11-211-4/+5
|
* move many functions from urpm.pm to urpm/media.pmPascal Rigaux2006-11-211-0/+1800