Re: ia64 added to svn
On Wed, Jul 21, 2004 at 10:24:24AM +0900, Horms wrote:
> On Tue, Jul 20, 2004 at 06:04:33PM +0200, Sven Luther wrote:
> > On Tue, Jul 20, 2004 at 01:16:36PM +0100, Martin Michlmayr wrote:
> > > * Matthew Wilcox <willy@debian.org> [2004-07-20 12:17]:
> > > > and myself are here; not sure about any other Debian kernel types.
> > > > What exactly do we want a policy on? Determining who should have
> > > > commit access to the kernel SVN repository?
> > >
> > > Who should have SVN access, how patch reviews are being done. I'd
> > > also like to see a policy clearly describing which kind of patches are
> > > suitable for the Debian tree.
> >
> > I would say the following :
> >
> > Apart from the current team, we need at least one person from each port (even
> > controversial port like amd64 seems to be) to have SVN access. SVN makes it
> > easy enough to roll out changes, and know who did do them, so this is no major
> > problem. Also, i guess it is not possible to regulate who has access to which
> > SVN area, so there is not much we can do, but we could use an rule of thumb
> > that those port maintainers should limit themselves to the part of the repo
> > relative to that port.
>
> Another issue I would like to see addressed at some stage,
> though it isn't really a policy issue, is what port/kernel version
> we support. By kernel version I mean 2.4 and 2.6. For instance
> the recent discussion of Bug#257001 suggets that m68k may be
> a candidate for not having a 2.4 kernel that is supported by this team
Well, do they have a 2.6 kernel then ? I suppose not, and that means only a
2.2 kernel.
I am myself doubtfull about 2.4 powerpc kernels, apart from the discover
issue, all would be nice and fine in 2.6 kernels. And i doubt it is making
much sense to have powerpc/apus support in both 2.4 and 2.6 as it doesn't even
work.
> (note I didn't say it won't, just that it is a possibility that is
> being discussed). I am concerned that in some cases there
> is just too little interest and too much work to make the package
> maintainable. Especially as in the case of the kernel one
> really needs access to the hardware to test the kernel -
> I can rebuild an m68k kernel on crest (very slowly) but
> I can't test it to see if it works.
Yes, so policy would be for those with the hardware to make the work. Helped
by us naturally. Nothing else really makes much sense.
Friendly,
Sven Luther
Reply to: