Re: Debian 2.2 Release.
On Tue, Jun 27, 2000 at 07:12:19PM -0700, Joey Hess wrote:
> Tomasz Wegrzanowski wrote:
> > On Wed, Jun 28, 2000 at 08:08:27AM +1000, Hamish Moffatt wrote:
> > > Now read my post. A heavily developed package with its own
> > > release cycle is never going to be up to date in our stable
> > > release, for various reasons.
> >
> > There is NO such reason.
> >
> > If they say they have new STABLE, we should put it into
> > both frozen and unstable as soon as debian/* works.
> > And if they support debian/* we should put it
> > into archives even in day when they announced stable.
>
> Naive. Very naive.
>
> Upstream authors may consider some degree of testing to be "stable".
> Debian may consider another degree of testing to be necessary before
> something is stable. It's just a word. Definitions vary. We can't expect
> upstream authors to have a definition that is the same as ours.
>
> Even if upstream authors do agree and follow 100% our definition of stable,
> we cannot just drop large masses of new code into a frozen distribution
> and expect there to be no problems. Making a distribution is all about
> integration; even perfectly stable new upstream code can require large
> amounts of integration to make it work with the rest of the system.
>
> If we actually followed your plan, I'm afraid debian would *never* be
> stable.
I was talking about applications (see my list : mozilla, freeciv, gnumeric...).
They usually can't break anything but themself.
I never said we should 100%ly trust stability of new
kernel, exim, pppd, glibc or gcc.
Reply to: