Re: Bits from the Release Team - Kicking off Wheezy
Joey,
nice to see you agreeing. :)
On 2011-04-28, Joey Hess <joeyh@debian.org> wrote:
> To most users of testing, a 5 month period when it doesn't update as
> much, but is also more constantly usable is mostly a draw; that period
> is when testing has the most new users.
That's also my observation. Users switch to it and test if everything's
still working and mostly stick to it. You get much more current
software than in stable and if there are issues you try to get them
ironed out pre-release.
So you get some users from the last stable release and the users that use
testing as a rolling distribution. With the frozen as present it would
be split into two camps.
One point seems to be that even if the freeze takes six months (which
would be ok, others[0] also release every six months and you're not
necessarily going to pre-releases there) you might end up with bits that
are actually much older because they were frozen well ahead (Iceweasel
being an example).
That might be annoying. It could possibly be countered with more
careful planning and perhaps more eager switching and backout plans.
But then a suite doesn't seem too agile to allow easy reverts.[1]
Improvement to update testing more quickly by easing the pain of
transitions, like rebuilding everything in a self-contained way
to avoid entanglements, would be well appreciated. The current plans
feel like a PR stunt to me, given that we won't notice it for quite
some time anyway.
What we do need is a testing that's constantly installable with d-i.
The unofficial d-i snapshots probably help with that. I think those
are efforts well spent. That said I think d-i itself should release
more often, too.
Kind regards
Philipp Kern
[0] GNOME, KDE, Ubuntu, you name it…
[1] 2.30+really2.28 or epochs. And a reverse transition with those
versions even breaking all our assumptions about shlibs
compatibility.
Reply to: