Re: Release process thoughts

[Lars Clausen <lrclause cs uiuc edu>]

On 21 Oct 2003, Cyrille Chepelov wrote:
> Le Tue, Oct 21, 2003, Ã 12:46:55AM -0500, Lars Clausen a Ãcrit:
> > On 20 Oct 2003, Tad Marko wrote:
> > > Perhaps frequent n.nn-rc-n releases? Mozilla and the Linux kernel seem
> > > to make good use of the release candidate (rc) release system.
> >
> > Well, that's basically what the preN releases are.  Problem is that in
> > order to end up with a stable release, the preN series can only accept
> > bug fixes (and non-coding patches like man pages and such).  So while in
> > preN, general development is halted, which makes us reluctant to go into
> > preN, thus creating greater gaps between releases.  If preN was
> > concurrent with further development, we could go into preN when there's
> > anything interesting, without delaying other development.
>
> Looks like you're trying to introduce Debian's "unstable/testing/stable"
> scheme, but at the source level (and with considerably less manpower).
> Given the less than impressive track record of Debian Testing, I'm not
> sure I'm excited by the prospect of going down that road.

Well, we're not taking on thousands of packages here.  I'm thinking more
like the kernel split in even and odd minor versions, but at a very small
scale.  Since most of the fixes that are found in the prereleases would
also need to go into the development branch, it wouldn't be much more work.
There is the risk that nobody gets around to fixing those bugs that need
fixing in the prereleases, which is why I would like somebody to be
officially in charge of that.  Basically, I would like the release process
to not hold up general development.

-Lars

-- 
Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause)| HÃrdgrim of Numenor
"I do not agree with a word that you say, but I   |----------------------------
will defend to the death your right to say it."   | Where are we going, and
    --Evelyn Beatrice Hall paraphrasing Voltaire  | what's with the handbasket?