Re: The nature of unstable (was: Danger Will Robinson! Danger!)
Paul M Sargent wrote:
>
> On a side note. I'm really not sure that this 'release' stuff works on
> debian. Coordinating the development cycles of an infinite number of
> packages is impossible. What I would like to see is an unstable tree where
> all development is done. As packages reach maturity they 'graduate' to the
> stable tree. A snapshot of stable tree at any time works. The unstable tree
> just becomes a place for developers to share packages.
>
What happened to the package pools proposal? It's as if Debian developers
are suffering from amnesia. I guess the package pools, as an idea at least,
had found a significant appeal in this list. According to some form of that
proposal,
what you've mentioned and even better release flexibility would be possible.
>
> The key point is a continually evolving release. As has been said before,
> Debian isn't commercial. It doesn't have to behave like it is with releases.
>
>
With a proper package pools system, Debian will have decoupled its archive
structure from its logical structure. Well, not that innovative but a
*linux*.com
company would certainly push it as a technological advantage.
It would seem that the independence of releases from actual archive content
would lead to the possibility of avoiding the current
rock-stable-and-completely-outdated
stable and scary-and-top-notch-unstable release cycle. People could then
prepare debian releases at some point between the two extremes. What is more,
I think developers could organize themselves as >teams<, who try to maintain
sub-releases for sub-systems. I think that kind of a flexible, and *distributed*
release management is pretty open-ended. That seems to be the right way to
motivate 500+ people working on such a large thing.
> To make this work major changes would have to be coordinated, but there is
> no reason that a major Perl change has to impact a major X change. Base
> packages like libc become more tricky though.
>
Absolutely. While debian stable has a great QA, people *who do not run it on
their
spacecraft* will eventually yearn for the latest software. However, they will
find the
unstable distribution "dangling" to their disappointment. For instance, I've had
some
months of insanity due to a mysterious breakage of printing tools in potato. Few
would
then dare to make an upgrade to unstable.
That is, you would certainly like some consistency and reliability, but not at
the cost
of missing a huge proportion of good software out there. That's what
distribution is
all about, right? Giving people some choices.
>
> Paul
> --
> Paul Sargent
> mailto: Paul.Sargent@3Dlabs.com
__
exa
Eray Ozkural,
CS, Bilkent Univ.
Reply to: