* André Wöbbeking [Wed, 30 Nov 2005 21:47:14 +0100]:
(Josh has already replied, but I had already written this before
realizing he had, so sending anyway.)
> On Wednesday 30 November 2005 20:26, Josh Metzler wrote:
> > On Wednesday 30 November 2005 02:17 pm, Yves Glodt wrote:
> > > why not stop wasting resources and halt work on 3.4.3 and go with
> > > 3.5 immediately, anyway etch will come with 3.5 I guess... no?
> > The release team asked us not to do that. They want the g++
> > allocator transition to go quickly, which means not uploading major
> > new versions of large packages like KDE.
> But for the g++ allocator transition all KDE packages must be updated
> anyway. Now you want to update all KDE 3.4 packages to update them some
> days later again with 3.5 packages. Isn't that more work than to go to
> 3.5 directly?
Yes, it is more work, but it is also what helps Debian as a whole the
most. See, experience tells us that a major KDE version (x.y.0) always
produces a bit of trouble on unstable, no matter how much you prepare
it in e.g. Alioth. See e.g. Christopher's comments [1] about the
possibility of kdelibs breaking on several architectures: then, we'd
be holding up the allocator transition for all of KDE apps, until we
fixed it, and further delaying the date of start of other transitions
to come (python, freetype).
[1] http://lists.debian.org/debian-qt-kde/2005/11/msg00332.html
So, instead, you upload to unstable versions that are known to work,
and which are known to not make other packages FTBFS, which enables
you (one the library is transitioned in all arches) to use semi-automatic
mechanism to update the rest of packages (binary NMUs), knowing _in
advance_ that they'll work. This all ends up in the transition that
goes faster (well, in Debian terms, anyway; objectively, it's still
long).
(The above mentioned semi-automatic mechanisms could have been used
for KDE itself, just recompiling all of 3.4.2 against kdelibs4c2a. But
well, kdelibs itself takes a while to be ready everywhere (hppa is
hitting badly on us), so while we wait, we go and upload 3.4.3, which
is more effort but it'll be nice to have in testing during all the
weeks that KDE 3.5 will be getting in shape in unstable.)
Hope this helps you understand better the decision. Check, specially,
the "and further delaying..." line in the first paragraph. :)
Cheers,
--
Adeodato Simó dato at net.com.org.es
Debian Developer adeodato at debian.org
From the moment I picked your book up until I put it down I was
convulsed with laughter. Some day I intend reading it.
-- Groucho Marx
