Re: How do I help?
On Fri, 9 Dec 2005, Nick Rusnov wrote:
How do I help is a question I often ask myself when I think of the sparc
port. There is a real lack of love for this port on various irc channels
and elsewhere because of a perception that the general quality is going
down, including endless troubles for newbies trying to install this
port. So, how can I help make it better? I have a few ultrasparcs (UE2s)
and a few sparc32s (various, some javastations, sparc classic, others).
I am a debian developer. I have a little extra time and try to help
newbies on the #debian-sparc channel all I can, but I feel like I don't
really know enough to be really helpful to those people with, for
example, installer problems. How do I help make sparc a better port for
everyone? Or is it really dying like people seem to think it is?
Hi Nick,
I don't think it's dying, but there are a few areas which require
attention. It looks like thanks to the efforts of Ben Collins and others
we are through with the buildd crisis (current stats on buildd.d.o
indicate that 98% of the archive is built). However, another problem is
the lack of developer-accessible machines, and I plan to start a separate
thread about it soon. Otherwise everything looks ok, you can find the
summary on the sparc Etch release certification page [0]. By the way, if
you are interested on working on sparc, you should definitely add yourself
to the list of developers on that page.
Here are a few ideas:
1. Porter work. This has been discussed recently, have a look at the
thread starting at [1]. One needs to examine build failures, evaluate
whether porting is feasible, fix bugs, NMU things if necessary, notify QA
team or file a bug for removal if the package is excessively buggy and
unlikely to be fixed. Here's what I've done during the last few days on
that field, as an example:
palo: pinged maintainer about bug 320284, ftbfs on sparc. Does not look
like it should be even attempted to build here.
wmkbd: filed 342917, package is in pretty bad shape.
icon: fixed (well, sort of) ftbfs on sparc (322972), it might be
considered for NMU since maintainer is MIA for quite some time.
cursel: filed a bug for ftp.d.o for removal (342637) as the build-dep
of this package has been removed from sid a while ago. Maintainer
contacted me saying that he does not object.
2. Test the installer on all kinds of hardware. The problems there are
usually trivial (except for the infamous 'Data Access MMU miss' and
similar stuff, which I was never able to reproduce on hardware I own),
but hard to solve without access to a sparc box. Successful installation
reports should be added to [0] as well.
3. Work on sparc32. Builds of sparc32 kernels are currently disabled in
linux-2.6. I've tried building 2.6.14 at some point, and it worked on my
SS20. If you would be willing to do regular testing of kernels and
installer, we could try and resurrect it, first by reenabling the sparc32
builds in the experimental kernels, then re-adding support for it to the
installer.
[0] http://lists.debian.org/debian-sparc/2005/12/msg00016.html
[1] http://wiki.debian.org/sparcEtchReleaseRecertification
Best regards,
Jurij Smakov jurij@wooyd.org
Key: http://www.wooyd.org/pgpkey/ KeyID: C99E03CC
Reply to: