[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Gnome to be removed from debian?



On Fri, Feb 12, 1999 at 07:51:10PM -0500, Ben Collins wrote:
> > Well, once the freeze is enforced on potato, it will be a snap to
> > settle the GNOME apps down. Release manager will just say: "Okay,
> > boys and girls, we now have gnome-libs 1.x.y, recompile with it.
> > Anything else will be treated as a release critical bug."
> 
> This kind of thinking is what slows down releases and prolongs freezes.
> You cannot just dump anything into unstable and figure that if it's
> broke it will be fixed at freeze time. The general consensus _should_
> be that what you put into unstable, you do as if it were in freeze, and
> plan on it being frozen at any time so we don't have these problems.

I was not referring to the explicitly broken package dependencies,
but on those that are not defined: for example, I have packages in
unstable that depend on libgtk 1.1.x. Which libgtk should I choose?

Normally one would think that the latest is the only one - but no,
instead we have ten of them. None older than two months (well, I
don't know exactly, but they're all rather new). And if I do pick
the current one, I'll have to recompile in three days.

To pick any of these would be right and wrong - right because noone
said you can't compile against some - and wrong because I have 10:1
chances that the one I'll be compiling on would not satisfy most of
the users. If not today, tommorrow.

To the current libgtk* dependencies there is no real solution. If the
final 1.2 version doesn't come out by potato freeze, we'll do what
we've done with slink - just take the current (then it was libgtk1.1_1.1.3)
version and make it 'the right one', and request recompilation of the
packages depending on other libgtk* versions. Recompiling to fix a
problem is easy - constant recompiling not fixing the real problem is
simply unacceptable.

(I hope libgtk* maintainer won't be offended by any of this, I know
it is vis maior.)

> Unstable is not a package dumping ground, look at the glibc 2.1
> packages, which are not even going to be put into unstable until they are
> tested.

No, I didn't say that. project/experimental exists.

--
enJoy -*/\*- http://jagor.srce.hr/~jrodin/


Reply to: