Re: ia64 added to svn
On Thu, Jul 22, 2004 at 03:06:35PM +0900, Horms wrote:
> On Wed, Jul 21, 2004 at 03:49:18PM +0200, Sven Luther wrote:
> > On Wed, Jul 21, 2004 at 10:38:29PM +0900, Horms wrote:
> > > On Wed, Jul 21, 2004 at 02:30:50PM +0200, Sven Luther wrote:
> > > > We remove the package from the archive ? I think that the right way for making
> > > > this claim is to see what debian-installer installs by default, or to have a
> > > > kernel-image-<port> which points at the default kernel.
> > >
> > > Should the Architecture of kernel source packages be set accordinly too?
> > > I suspect not.
> >
> > No idea what you mean, but i suspect this is not necessarily. The
> > kernel-source package is arch: all after all.
>
> Let me try and clarify.
>
> The kernel-source-2.4.26 has
>
> Architecture: source all
>
> To me this implies that it is reasonable to expect that it will
> compile on all ports. But in the case of m68k (and possibly others)
No, this is not, and has never been the case. This only means that it is an
arch independent package, and thus there is only one binary package for all
arches. Furthermore, the actual source tarball is usefull on all arches, even
if it doesn't build, or work or whatever, be it only for those working on
fixing it, or applying a patch (like a port patch) or coing cross compilation
(and i used to do m68k->ppc cross compilation back then :)
> this is not the case. And as I have concluded from discussion elsewhere
> on this list, is unlikely to be resolved in the forseeable future.
So, there is no kernel-image or kernel-patch package for said kernel
version/architecture, and that's it.
> My question is, do we have a way to tell a user, this
> is a source package, go for your life, but we know
> you are going to have problems on m68k (and...).
No, not really, it has long been known that a given source package needed
patches. What could be done is add in the description the list of subarches
which should work out of the box.
> Or is it enough that it is a source package and implicitly
> may or may not be able to be used out of the box to produce
> something on architecture x, y and z?
See above.
> What I am really getting at is the next time an m68k user
> downloads kernel-source-2.4.26 and tries to build a kernel from
> it, it will likely fail for them too. It seems reasonable that
> they would then open up a bug. Which we will probably then mark
> as wontfix. It would be nice to make it more obvious to the
> end users that it isn't going to work form the outset.
Like said, all non-x86 subarches needed to apply patches anyway, so ...
Friendly,
Sven Luther
Reply to: