Bug#729289: transition: openscenegraph
2014-02-02 Niels Thykier <niels@thykier.net>:
> On 2014-02-01 14:11, Rebecca N. Palmer wrote:
>> The state of openscenegraph itself isn't the problem here: what we're
>> waiting for is for simgear to clear NEW and for Release Team to give
>> official permission (britney hint??) to ignore openwalnut for now.
>
> Actually both of those things are FTP master decisions. If openwalnut
> had been in testing, then that part would need both FTP masters and the
> Release Team - but since it is not, then we just need the FTP masters
> approval for getting openscenegraph decrufted at the price of breaking
> openwalnut (even further).
>
> I can ask the FTP masters to process simgear - decrufting openscenegraph
> will have to wait until we have fixed all the packages that we can.
>
>> There is the separate question of whether you want an ~rc in an Ubuntu
>> LTS (14.04 freezes 20 Feb), but they finished their transition to
>> 3.2.0rc months ago (by removing openwalnut) and will sync from
>> experimental if you ask them to, so that isn't tied to what we do in
>> Debian.
>
> Based on Manuel's mail (not quoted here), I think I would prefer if we
> transitioned to openscenegraph ABI 99 rather than 100. In my eyes, it
> makes the whole thing simpler and means we can start working immediately
> - especially considering that our best guess on a release date for the
> new version is more than a week or two away (Manuel suggested March or
> later).
Actually, I meant to say that we (or at least I) don't have any
confidence in the accuracy of our best guesses for the date when 3.2.1
final will be available (meaning: I don't have much confidence in
upstream's predicitons, or their assurance to not change ABI again for
3.2.1 final). Because the estimations from upstream have been "any
time now" for months (repeated every now and then), but with a delay
of more than 3 months since the initial estimate. But it can also be
released in the next days or weeks... nothing prevents that, AFAIK.
Perhaps Alberto or Rebecca (who follow upstream mailing lists) have
better guesses about the current state of mind of upstream.
So we finally gave up on waiting and fixed the current version in
Debian unstable in mid January without waiting any longer, trying not
to continue causing disruption in Debian. From my part, and I think
Alberto's, any solution minimising problems for the rest of Debian
will be fine.
Given that any newer upstream release (even 3.2.0 final) will have to
go through the NEW queue (ABI/soname changes), even if they release
tomorrow it will take a while to get to unstable anyway; and then
there's the round of binNMUs and NMUs. So for the sake of rdeps, we
thought that stabilising with ABI 99 is a good idea, or else we might
continue in the same situation for months.
Also, for the future, question to Niels: we know that multiple
versions of the same library are discouraged, but maybe it would be
useful in this case to accomodate to the pace of different rdeps?
This library doesn't exist in Debian only (or perhaps even mainly)
because the rdeps in Debian, but because of the library itself, and
many people use it independently of other packages installed. That's
why it is desirable to have the latest versions quickly as well.
Cheers.
--
Manuel A. Fernandez Montecelo <manuel.montezelo@gmail.com>
Reply to: