Re: Sarge TODO items
On Wed, Jun 02, 2004 at 12:30:45PM +0200, Christoph Hellwig wrote:
> We've been flaming^H^H^H^H^H^H^Hdiscussing a lot of design items here,
> but we should try to get up a list of items for to be done for sarge.
>
> First priority for all sarge work should be to keep the maintaince
> overhead for the security team down. That means in particular trying
> to keep as few as possible sourcebases around the need fixes applied to
> build new images.
>
> * Linux 2.2
> - I'd really really love to avoid shipping 2.2 for sarge, but it
> seems that's not possible :P
> - update from 2.2.25 to 2.2.26 to include upstream fixes
> - try to get as much as possible of the intrusive arch patches
> into kernel-source to at least have a single 2.2 codebase to patch
>
> * Linux 2.4
> - move everyone up to 2.4.26 or better 2.4.27-rc
> - remove intrusive patch that are provided in 2.6 (ipsec or ea/acl)
> - try to merge arch patches into kernel-source as much as possible
I have been tracking the -benh tree for the powerpc 2.4 kernels, since
most people used that anyway. Benh has not updated this tree since
february though, and i don't forsee him working on this in the next two
month or so, so there is no 2.4.26 powerpc kernel easily possible.
Furthermore, the integration of those arch specific patches would be
difficult, as the -benh tree is not against the linus bitkeeper tree,
but from an older powerpc bitkeeper tree. I don't really believe working
on doing this integration is worth it for 2.4 kernels on powerpc, better
make all the possible so we don't really need 2.4 kernels anymore or
something such. I had not planed to release a 2.4.26 or later package
for powerpc under these conditions.
> * Linux 2.6
> - change the kernel-source packaging so it nicely deals with lots of
> little patches (can some with enough packaing fu help out on
> this?)
> - merge up to 2.6.7(-rc) to pick up all the stuff that went upstream
> and reduce the diff vs upstream nicely
I would strongly discourage you to start tracking -rc kernel sources,
staying with full releases is better, i think.
> - try to merge patches from the architecture maintainers into
> kernel-source as much as possible.
Yes, but without making this top priority for sarge. Sarge+1 is another
story.
> To do all this work nicely we should get a repo on alioth ASAP,
> especially to allow the arch maintainers to work on the main
> kernel-source too for better cooperation.
Yep. Friendly,
Sven Luther
Reply to: