Re: Linux 2.6.8-rc1

[Paolo Ciarrocchi <paolo.ciarrocchi@gmail.com>]

On Mon, 12 Jul 2004 18:34:17 +0200, Adrian Bunk <bunk@fs.tum.de> wrote:
> On Mon, Jul 12, 2004 at 05:56:14PM +0200, Paolo Ciarrocchi wrote:
> >...
> > > OSDL does some tests for any -rc and many other people like me do other
> > > testing. Besides this, most patches already got similar treatment in
> > > -mm. This might not be a base for an ISO 9000 certificate, but it seems
> > > to be sufficietely working for finding most problems before the acttual
> > > release.
> >
> > OSDL does some test for any -rc but the results of these tests don't affect
> > the release process. At least not in an official way.
> 
> The Linux kernel development process isn't that much formalized. But if
> someone finds a serious new problem in a -rc kernel a fix will usually
> go into the next -rc.

I agree.

> Compared to some other open source projects like e.g. Debian the Linux
> kernel has a pretty well-working release process (and the 2.6
> development avoided several mistakes of the 2.4 development).

I agree, again. 

> > > It would be more important if Linus would release one last -rc that will
> > > be released unchanged (except for EXTRAVERSION a few days later to catch
> > > bugs in last minute changes. This might catch more problems like the JFS
> > > compile problem in 2.6.7.
> >
> > Right,
> > and in those days may be OSDL could run the testsuite we are discussing about.
> 
> The way kernel releases currently work IMHO works well with the
> exception that there should be a last -rc that should be released as
> -final a few days later if no problems show up.
> 
> What other actual problems do you currently observe?

I really don't want to say that the process is now not working,
I'm just saying that it could be improved.

I totally agree with you when you say that there should be a last -rc
that should be released as final if no problem show up.
What I'd like to add is that in the few days between the last -rc and the final
there is enough time automatically run a few tests like the ltp suite,
the compile
stats and a few regression test.

If nothing strange show up, both from the comunity and the automatically tests,
the new version is released.

I don't want to reinvent the wheel, I see that all the tools are
already here, most of them are maintained by OSDL, so *may be* we
should formalized a bit more
the release process.

I don't see any disadvantage in doing this.

ciao, 
Paolo

-- 
paoloc.doesntexist.org