[Aptitude-devel] Bug#488081: aptitude: prefers to take from experimental than unstable

[Manuel A. Fernandez Montecelo]

Control: tags -1 + moreinfo


Hi,

2008-06-26 14:49 Daniel Burrows:
>On Thu, Jun 26, 2008 at 10:24:58AM +0200, Yann Dirson <dirson at debian.org> was heard to say:
>> Package: aptitude
>> Version: 0.4.11.4-1, 0.4.11.6-1
>>
>> With today's updates, sid's wengophone requires "libgnutls26 (>=
>> 2.4.0-0)", not satisfied by lenny, but by unstable (2.4.0-2) and
>> experimental (2.4.0-1), which poses a problem to aptitude after selecting
>> everything for upgrade.  In that case, for some strange reason the
>> packages from experimental are considered better than the more recent ones
>> in unstable !  Note the identical aptitude scores, despite apt assigning
>> correct scores to the versions:
>>
>> # LC_ALL=C apt-cache policy libgnutls26
>> libgnutls26:
>>   Installed: 2.2.5-1
>>   Candidate: 2.2.5-1
>>   Version table:
>>      2.4.0-2 0
>>         500 http://ftp.fr.debian.org unstable/main Packages
>>      2.4.0-1 0
>>           1 http://ftp.debian.org ../project/experimental/main Packages
>>  *** 2.2.5-1 0
>>         990 http://ftp.fr.debian.org testing/main Packages
>>         100 /var/lib/dpkg/status
>
>  aptitude considers all "extra" (i.e., non-default) versions to be equal
>when scoring dependency solutions.  That's probably what's happening
>here.

I think that the behaviour has been changed since the version of the
original report, and aptitude better respects priorities set in
apt_preferences (lots of changes for 0.6.* and some for 0.7.* series).

Did you observe similar problems lately?  I haven't observed them myself
even if occasionally mixing release suites (inc. experimental), nor saw
recent reports about similar problems (but hidden among the still
hundreds of open bug reports, who knows).


The rest of the discussion in the bug report is a bit out of scope with
respect to the the original report.  Even if the underlying issue wasn't
fixed while this bug report stayed behind not closed (which is what I
think that it happened), I don't think that there's much that we can do
by now to guess what the original maintainer had in mind.


Cheers.
-- 
Manuel A. Fernandez Montecelo <manuel.montezelo at gmail.com>