Re: Draft release notes and errata for RC1 - please comment
On Tue, Nov 14, 2006 at 05:55:54AM +0100, Frans Pop wrote:
> On Monday 13 November 2006 13:44, Sven Luther wrote:
> > > You did not take into account my comments, there is no such thing as
> > > an "OldWorld PowerPC". The correct name is "oldworld powermac".
> > > Please try to be exact.
>
> I did not get that from your previous mail.
> Probably because it was somewhat buried in the text. I've said often
> enough that I'm not really familiar with powerpc, so excuse me if I miss
> the fine distinction between powerpc and powermac.
Well i listed you all the existing arches and mentioned this. If you had not
been overly sensible over the "abysmaly confusing" words, maybe you would not
have missed those.
> Fixed now.
>
> > > You can add :
> > >
> > > * IBM pseries on serial port : finish-install does not set the
> > > /etc/inittab serial console. You need to edit /etc/inittab and set
> > > your serial console by hand. (See #394970 for details).
> > >
> > > * Apple G5 fan will go into full speed mode.
> > > (See #392921, #394971 and #394972 for details).
>
> Don't think these are significant enough that they need to be mentioned.
Sure they are. the first one will mean that users will facing a serial console
with no getty waiting on it, and will think that the installtion went bad
(been there, users get really confused, even i needed some time to catch the
trick). Fixing this by chroting in the /target and adding the line by hand is
trivial, so putting this in the errata would have been of uttermost
importance.
Furthermore, the fan issue is primordial. The G5 machines are *UNUSABLE*
without earplugs when fans go into full gear. I got many reports from users
about this.
> > Oh, and :
> >
> > * RAID installation is broken with macintosh partition tables.
> > (See #397973 and #392767 for details)
> >
> > * PReP unsupported for RC1 due to missing support in the 2.6.17
> > kernel. Support is present in 2.6.18 as a separate -prep flavour.
>
> Added.
Cool.
Just one note about powerpc, so you are a bit more cluefull next time :
dead powerpc subarches :
- apple nubus machines : never supported, need 2.4 kernels.
- apple oldworld : support a bi broken, but should basically work.
- apple newworld : support is nice, but they aren't produced anymore.
- apple G5 boxes : same as newworld, use -powerpc64 variant.
- amiga/apus : support dropped, few users left, patches need to be
upgraded to recent kernels.
- PReP : boxes from motorola, IBM, Bull and others.
- 32bit CHRP from IBM : i don' t believe those are produced anymore, they
may be supported though, should work well.
- 32bit CHRP from Genesi (Pegasos) : due to rohs regulation, production is
stopped, butsee below).
Live powerpc subarches :
- IBM 64bit chrps (pseries, iseries, blades, ...) : those are very much
active, but the support in d-i is not so good. I had access to a js21
blade center in july, and was more recently lent a p505.
- Genesi new 32bit models : efika and MPC8641D based boards. They behave
mostly like the pegasos as far as d-i is concerned. Need some patches
and stuff.
- Mercury ppc970 (non-apple name for the G5) 1U servers: never seen one of
those, not sure what the support is and if they are chrp, yellowdog
should run on them.
- sony playstation 3. Patches are out there, i was asked by sony to do the
integration, but no hardware is available. Would be nice to have support
in etch though :)
- assorted random powerpc based consoles : the gamecube, nintendo wii and
xbox 360 are all powerpc based. No idea if they are supportable.
- assorted embedded devices : routers, synology NAS, i think the kurobox
too. No support in debian. Not sure if it would be easily doable. These
are NLSU2 class thingies. There are other powerpc devices which are less
visible, running your car or stuff like that. Not sure if we want debian
in them though :)
As you see, with apple going to intel, none of the still produced powerpc
machines is really all that well supported in d-i, and thus, without support
for the IBM boxes, new Genesi models, and maybe the PS3, powerpc would fail
the RM's criterias for a release arch.
I have been working hardly to make at least good IBM support possible, so
please don' t put stops into this by delaying patches unduly, and i ask you
again to reconsider both my svn commit access, as well as for example allowing
Aurelien Gerome commit rights, since he is working on miboot, emile and
yaboot, and could give Colin a hand with yaboot-installer (i discovered a
yaboot-installer/raid1 bug yesterday for example, need to reconfirm, with a
new install before the box goes into production, and i will file a bug report,
but i am useless to provide patches on either yaboot-installer or partman).
Friendly,
Sven Luther
Reply to: