Re: debian 2.1r3 needed and release team coordination
Adam Di Carlo wrote:
> IMHO, we should be thinking about Debian 2.1r3. This is what I think
> needs to be in Debian 2.1r3 and why we need it:
>
> * replace kernel-source 2.2.1 pkg with 2.2.5 or better
> Rationale: 2.2.1 is very evil
Hmm, 2.2.1 or 2.2.5, it was "use at your own risk". If 2.2.1 has known
problems then it would be more consistent with our policy for stable
to simply remove it.
> * any architectures based on 2.2.x (sparc/Sun4u is the only one I
> know of) should be be 2.2.5 based or better
> Rationale: see above
... but if this is so, then we can't. Are you sure that there are
architectures in slink that use a 2.2.x kernel? linux 2.2 wasn't
even released when we froze sparc.
> * utterly broken packages and things should be fixed -- jadetex is
> broken (jadetex for which I shall NMU for stable soon); there are
> a number of things for Debian/sparc that I know of that are broken
> (setserial which hardcodes IRQs, permissions on /dev/fd* are
> wrong, /dev/mouse doesn't exist)
> Rationale: we're probably talking about < 10 packages where we
> need to identify the problems and fix them; no new upstream
> sources required here AFAIK)
Hmm, except for setserial, these are bootfloppies problems, aren't they?
> * kernel-image 2.0.36 on i386 has very broken Adaptec (namely,
> AIC7xxx) support
> Rationale: boot-floppies is getting flooded with bug reports about
> this
Okay, both of these points are about grave bugs. I'm okay with the
idea of relaxing "security bugs only" to include grave bugs, but then
we really need a more solid structure for such changes. That means
explicit testing, and the packages should be built for all architectures
that we support. We need this for security updates as well. Can
it be done?
> * we're ready with boot-floppies for Croatian and French versions; I
> bet slink-cd isn't yet ready, but even so it would be nice to
> release these
> Rationale: try to start momentum for multi-lingual Debian, which
> is a major goal of potato, IMHO
I don't think this is a good idea. As you said, it's something for
potato. We should start work on potato bootfloppies; let's not
involve the slink bootfloppies in this work.
> What I'd like to do is first of all, get consensus on whether we need
> a point release of slink, and what needs to be in it. My stuff above
> is a starting point.
Point releases for slink are already being made for security updates.
I think we have to consider other changes on their individual merits.
Richard Braakman
Reply to: