Re: heads up

[Michael Banck <mbanck@debian.org>]

Hi,

On Wed, Mar 16, 2005 at 11:04:39AM -0800, Thomas Bushnell BSG wrote:
> Some changes to the way Debian manages its archive and releases
> architectures may be in the works.  Most of this doesn't affect Debian
> Hurd because we aren't a release architecture anyway.

To be precise, we were a 'second class citizen' for all the time.  So in
case we manage to stay at that level for the time being, we might even
profit from it.  E.g. unstable snapshots look very much like what Philip
Charles did with the CD releases in the past, but we could have the
packages in the archive proper.  Also, other to-be-scc'd arches will
have to implement their own testing and release infrastructue, so we
could borrow from that.

But note that this is just a proposal and things might change (and they
are changing)

> Our arch is ordinary i386 hardware, so we meet this.
> 
>   2. the architecture must be able to run a buildd 24/7 sustained
>      (without crashing)
>   3. the architecture must have an actual, running, working buildd
> 
> This, I think needs to be our target.

I have talked to Martin Michlmayr yesterday and he promised support for
a autobuilder box and another Debian Developer has already stepped up
and volunteered hardware.  I will talk to the local admins at my
university about hosting a box (I got a low-performance one already)
there tomorrow.  There are other alternatives like e.g. the HurdFr
boxes, but we have not talked to HurdFr officially about how this would
work out.

>   4. the port must include basic unix functionality, e.g resolving
>      DNS names and firewalling
> 
> We certainly do DNS names, and I'm asking what specific firewalling
> support is required.

When confronted with this requirement, the people who made them said
they just want to keep out 'toy architectures' from the archive.
Whether this applies to us I don't know, different people have different
opinions.

There is probably not going to be a problem with the other ones, we have
enough DDs and users, we just need to get packages compiled. Right now,
getting core packages ported was our main concern, but we will have to
switch to getting as many packages currently built as possible for the
time being.  The rationale (i.e. we cause a lot of old source packages
being left in the archives as we have outdated binary packages) makes
sense.

I propose to let the dust around this proposal settle and then talk to
the ftp-masters (as those appear to be responsible for the
'in-the-archive-at-all' bit, while the release managers decide who gets
to be a releasing arch; we're not that far yet) calmly about what needs
to be done.


cheers,

Michael

-- 
Michael Banck
Debian Developer
mbanck@debian.org
http://www.advogato.org/person/mbanck/diary.html