Re: re evolution
On Tuesday 04 April 2006 09:13, hendrik@topoi.pooq.com wrote:
thanks all for chiming in. This is the first time i really wanted to try a
test distro, so I was only trying to do my part and report something that
looked broke. In installing etch i actually expected to find more things
that needed tweeking. I have been very happy with what I have found and
would like to further be involved.
thanks
caleb
> On Tue, Apr 04, 2006 at 01:44:23PM +0100, Nikolay Kichukov wrote:
> > hello,
> > Just to make sure:
> > Is the testing branch the one that comes after
> > unstable, or is it the other way round?
>
> Here's my understanding:
>
> Packages are allegedly debugged (almost no one can *really* debug, as
> testified by security upgrades) and places into unstable/sid. Since the
> developers mostly (I melieve) build and test their packages on
> unstable/sid, we can expect their declared dependencies to be satisfied
> in sid. But the packages are sometimes not fully debugged and sometimes
> just fail to work properly. So the package manager will have little to
> complain about, but the user might.
>
> > I had many discussions with friends, and it seems
> > noone can say for sure which is the most rapidly
> > changing distro.
> >
> > As far as I know, testing is not so oftenly changing,
> > and the code is believed to be somehow more "stable"
> > than the experimental unstable. That is why packages
> > from testing are getting into stable after some time.
> >
> > Then, why is it, very often, it happens to
> > applications to totally "stuck" in the testing branch
> > after dist-upgrade?
> >
> > It happened with evolution now, and it has happened to
> > other apps before.
>
> Once packages look good enough, and all their dependencies do too,
> they are
> moved to testing. So more debugging will have been done, and the users
> will have fewer conplaints. But the dependencies (either what the
> package depends on or what other packages depend on it) may not all have
> been checked properly (after all, lots of stuff has different
> versions between testing and sid), so the package manager will have more
> complaints.
>
> On occcasion, packages have been promoted to testing even though they
> were known to break dependencies. This happened last fall in the
> transition to xorg and the new c library -- waiting until *every*
> obscure package had been recompiled and stabilized for testing would
> have unduly delayed the main sunsystems like KDE and gnome. So I
> read on the mailing list, anyway.
>
> Once in a blue moon or so, an entire distribution is deemed to be
> 'stable', replacing the previous stable distribution. This is
> apparently done by changing a few symbolic links in the package
> repositories, so no new incompatibilities arise within the distribution
> by making this change. So the newly-deemed-stable distribition really
> remains stable. But people who update their systems from 'stable'
> instead of from 'woody' or 'sarge' or 'etch' will find huge changes
> occuring at that time.
>
> > Thanks & Regards,
> > -Nikolay Kichukov
>
> You're welcome.
> -- hendrik
Reply to: