Fighting the i386 plague
Phil Knirsch
pknirsch at redhat.com
Wed Jan 24 09:30:38 UTC 2007
Bruno Wolff III wrote:
> On Tue, Jan 23, 2007 at 14:11:39 +0100,
> Dominik 'Rathann' Mierzejewski <D.Mierzejewski at icm.edu.pl> wrote:
>> If two packages own the same file then it's a bug in packaging!
>
> If the shared file is the same in both packages, this can be a simpler
> solution than creating a whole other package just to contain the one shared
> file.
>
So what you would be suggesting is to create new subpackages for any
possible shared data?
First of all, this would lead to a huge package inflation, especially
because of multilib packages. Also, this would have to be continually
monitored for all packages in order to be useful and would be a real
packaging nightmare as nearly any file from any package could be shared
basically. This is definitely not doable for 2200+ packages in Fedora
Core alone. Now add in the 4100+ Extras packages that will get reemerged
and you're looking at more than 6300 packages where each package can
share a file with another one and each change to each package filelist
needs to be tracked and monitored and possibly a new subpackage created.
Second of all, i don't think thats the right solution anyway. As has
previously been already stated, if 2 packages install the same file i
see no reason why during removal it shouldn't be possible to check
whether the same file is still "in use" by another package and then it's
just not removed. So a useful and simple algorithm already exists, so
why not use that.
I know file conflict/duplicate checks can be rather expensive
computational wise, but even with the current form of the rpmdb it's
possible to implement that rather efficiently.
Read ya, Phil
--
Philipp Knirsch | Tel.: +49-711-96437-470
Development | Fax.: +49-711-96437-111
Red Hat GmbH | Email: Phil Knirsch <phil at redhat.de>
Hauptstaetterstr. 58 | Web: http://www.redhat.de/
D-70178 Stuttgart
Kaa's Law: In any sufficiently large group of people most are idiots.
More information about the fedora-test-list
mailing list