* Marc Haber <mh+debian-de...@zugschlus.de> [100626 14:07]:
> On Fri, 25 Jun 2010 14:27:31 -0700, Steve Langasek <vor...@debian.org>
> >The footnote to Policy 3.5, where this is written out?
>
> Ah, so this is the same as the no-circular-dependency rule, dumping
> extra error proneness and extra thoughtweight on all developers
Please, try to be a bit more fair. Having people not need to specify
dependencies is really not the solution that "dumps extra error
proneness and extra thoughtweight" on the developers. And once those
need not there saying they should not be there is not adding any error
proneness.
I'm personally all in favor for making the "hard to deinstall"
and "not needed in dependencies" different things (and am annoyed
every time my build-chroot has to contain many unnecessary packages
just because Essential forces to install e2fsprogs and mount
and things like that), but have to acknowledge that while the current
situation is the less flexible one, it is also the less error prone one.
> to work around shortcomings in our software?
If you read the second paragraph it also gives a reason that has nothing
at all to do with making it easier for software [1]: If there are no
dependencies, essential stuff can just move between packages or have
packages renamed.
Not requiring dependencies but allowing them only combines the
disadvantages of both worlds. It does not really make sense to have
both except for transition periods.
Bernhard R. Link
[1] and thus more likely for our users to get smoth upgrades. I really
do not see the point why just because some software could be able
to solve something, we should choose the solution most likely to
break.
Especially circular dependencies are a pain whenever you have some
seriously broken system or are otherwise forced by whatever reason
to fall back to manual dpkg invocations....
--
To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Archive:
http://lists.debian.org/20100626154515.ga13...@pcpool00.mathematik.uni-freiburg.de
