On Thu, 30 Mar 2006, Frederik Schueler wrote:
Hello, I would like to schedule the upload of linux-2.6 2.6.15 for tomorrow, friday 31st. Thanks to Waldi, the external modules build issue is fixed now, making this a pretty important update. Does anyone have other issues left which should be in -5, or does the schedule fit? The other upload candidate is linux-kbuild-2.6: like kernel-kbuild-2.6-3 in sarge, this fixes the problem of cross-compiled linux-header packages containing binaries from the host-arch and not the target arch, as those binaries are shipped in this kbuild package.
Ok, I have a comment on that. Isn't one of the major goals of making kernels to build from a single source was to avoid the pain of having to deal with separate kbuild package? I consider it a major change, why wasn't it discussed on the list? I have absolutely no clue why such change is required, and I don't think other members of the team do either.
As you can probably tell, I am not very happy with the way the kernel team operates. In the last few weeks commits with legitimate bugfixes by myself and other people have been reverted without explanation. I have tried to start the discussion on what I consider an important issue - maintenance of out of tree modules - only to receive the feedback of 2 (two) members of the kernel team (out of registered 28) and a message from Bastian effectively saying that something is already implemented, and there is no need to bother. The bug list is growing, initramfs generators are constantly buggy and we still cannot make a major release without screwing up the linux-headers in one way or another.
What I would like to see is more attention to the quality assurance, and less to pushing new untested code into the archive, which leads to embarrassment for the team pretty much at every release. It's true, usually the problems get fixed fast, but at what cost? The first week after 2.6.16 we were uploading pretty much every day, and on slower arches the buildds were finishing the builds just in time for the next upload. Do our users benefit from that? I sincerely doubt it.
I don't know whether there is a solution for all these problems we are facing. I believe that having somebody in a position of at least nominal authority would help a lot. Such person could make sure that the design decisions are properly discussed, and would have the last word in the questions which are controversial, or which no one cares about. He could finally introduce a reasonable upload schedule, coordinate tasks, and care about QA issues a bit more than we currently do. He could plan ahead, having in mind that after all, we do plan to release at some point, and we have to stop hacking on the packaging and try to stabilize it.
So, that's my rant. I don't see a point on working on packaging anymore, since it is not inconceivable that at any point my work might be overwritten by somebody else. Please don't take it a threat to resign (or as a proposal to expell someone :-), I do enjoy working on sparc kernel and on general bugs. I would just like people to think about where we, as a team, want to be now and in the future.
Best regards, Jurij Smakov jurij@wooyd.org Key: http://www.wooyd.org/pgpkey/ KeyID: C99E03CC