[Aptitude-devel] Bug#799918: Fwd: Bug#799918: Bug#799918: apt-get proves I am innocent

[Manuel A. Fernandez Montecelo]

Forwarding message, the last few messages were not copied to the bug
and this one summarises them.


---------- Forwarded message ----------
From: Manuel A. Fernandez Montecelo <manuel.montezelo at gmail.com>
Date: 2015-10-06 17:15 GMT+01:00
Subject: Re: [Aptitude-devel] Bug#799918: Bug#799918: apt-get proves I
am innocent
To: 積丹尼 Dan Jacobson <jidanni at jidanni.org>
Cc: "799918 at bugs.debian.org Axel Beckert" <abe at debian.org>


2015-10-04 1:47 GMT+01:00 積丹尼 Dan Jacobson <jidanni at jidanni.org>:
>>>>>> "MAFM" == Manuel A Fernandez Montecelo <manuel.montezelo at gmail.com> writes:
>
> MAFM> It looks like one of the problems in the last lines is that qgis is
> MAFM> incompatible with the version that is to be installed by default from
> MAFM> ligbdal1i from experimental (do you have experimental with higher
> MAFM> priority?  it seems that it installed that by default).
>
> Yes,
> # apt-cache policy
> Package files:
>  100 /var/lib/dpkg/status
>      release a=now
>  500 http://ftp.tw.debian.org/debian unstable/non-free i386 Packages
>      release o=Debian,a=unstable,n=sid,l=Debian,c=non-free,b=i386
>      origin ftp.tw.debian.org
>  500 http://ftp.tw.debian.org/debian unstable/contrib i386 Packages
>      release o=Debian,a=unstable,n=sid,l=Debian,c=contrib,b=i386
>      origin ftp.tw.debian.org
>  500 http://ftp.tw.debian.org/debian unstable/main i386 Packages
>      release o=Debian,a=unstable,n=sid,l=Debian,c=main,b=i386
>      origin ftp.tw.debian.org
>  990 http://ftp.tw.debian.org/debian experimental/non-free i386 Packages
>      release o=Debian,a=experimental,n=experimental,l=Debian,c=non-free,b=i386
>      origin ftp.tw.debian.org
>  990 http://ftp.tw.debian.org/debian experimental/contrib i386 Packages
>      release o=Debian,a=experimental,n=experimental,l=Debian,c=contrib,b=i386
>      origin ftp.tw.debian.org
>  990 http://ftp.tw.debian.org/debian experimental/main i386 Packages
>      release o=Debian,a=experimental,n=experimental,l=Debian,c=main,b=i386
>      origin ftp.tw.debian.org
>  990 http://mozilla.debian.net experimental/iceweasel-aurora i386 Packages
>      release o=Debian Mozilla Team,a=experimental,n=experimental,l=Debian Mozilla Team,c=iceweasel-aurora,b=i386
>      origin mozilla.debian.net
> Pinned packages:
> #
>
> MAFM> Maybe it would have worked with no need to or less solutions offered if
> MAFM> you were using unstable.  That part is a package-specific issue, I think.
>
> I bet it would.
> However fortunately I didn't,
> else no one would have ever discovered this bug!

I think that somebody else has, more on that later.


> MAFM> On the package manager side of things, both apt and aptitude choke on
> MAFM> the fact that it's a virtual package or something, I don't understand,
> MAFM> and I cannot investigate immediately.  There are an awful lot of virtual
> MAFM> packages (~14K with multi-arch enabled, from the interactive interface),
> MAFM> even if many are not actually used / depended upon by any other package,
> MAFM> and could be pruned.  So it's strange that apt says that it cannot find
> MAFM> candidates for it while it must work for all those other packages.
>
> I don't know about all that. I'm just the simple bug reporter/tipster.
>
> MAFM> I also don't know why qgis depends on both libgdal1i and the virtual one
> MAFM> with the full version in the name.  In my system installs fine because
> MAFM> libgdal1i is already installed, because one package that coincidentially
> MAFM> I co-maintain depends on it.  My package only depends on libgdal1i, so
> MAFM> it had no trouble installing.
>
> Ah, but if it weren't already installed... the bug(?) will reveal
> itself.
>
> Anyway, soon the maintainers will probably change something, so the
> opportunity to fix this bug will be gone... only for it to strike again
> years later in some other package... So I would hope somebody would
> look into it soon.

I don't know if I mentioned before, but I have impression the that the
bug is a slightly different version of this one:

  https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=316633

and this as a possible duplicates more close to yours:

  https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=798898


That is, the virtual package cannot be installed for some reason, and
aptitude is not informative enough (appending something to "which is a
virtual package" like "but it is not installable because of
conflicts").  In the old bug report above, because there is no package
providing it, in your case and the other recent one because of
conflicts.

So, if I am right in the hypothesis above, we could improve the
message the main problem would remain, and that is that the packages
in their current state are not installable anyway, and there is
nothing that we can do from aptitude's side to get it installed in
your system.


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