Anyone have a clue how to handle the GStreamer thing?



I sure don't know how to handle this.  I've been thinking about it for
a while, and the best I can think of is sending an email like what
I've typed up below, and asking Thomas & Ronald to move the discussion
(*minus personal portions*) to d-d-l.  Anyone have any alternate
suggestions or improvements...I hope?  Given how late things already
are (feature & module freeze is supposed to be less than 24 hours
away), I'd love it if someone else took over in order to get this out
before I woke up.  *hint* *hint*.  ;-)  Otherwise, I'll check my email
in the morning and all the enlightened ideas that you _will_ have sent
me by then.  :-)




Hi everybody,

So, we messed up and didn't keep close enough tabs on everything,
resulting in discovering an issue pretty late.  This will probably
force a late decision -- one which will be even later because it's not
a clear cut decision and thus not one the release team should really
make; we need to try to find rough consensus in the community, if
possible.  Some background on the problem:
http://live.gnome.org/TwoPointThirteen_2fDesktop (and the jhbuild
modulesets as a result) listed GStreamer 0.8 to be shipped with Gnome
2.14.  Quite a few people were assuming that 0.10 was the plan for
2.14 and were totally unaware that 0.8 had even been on the plan. 
Ubuntu and Fedora development versions (i.e. the distros that I
checked or found out about) seem to both be headed towards 0.10.  I'm
not an expert on the issues, but from what I understand, additional
issues playing a role are that 0.8 will be nearly unmaintained _but_
0.10 has multiple regressions relative to 0.8 that are unlikely to be
fixed by release time.  Thomas and Ronald and others can flesh out the
technical details much better than we can, so we'll let them do so.

Now, it is possible to have both 0.8 and 0.10 installed at the same
time, so the decision could be to ship both and have modules use 0.10
if it works for them and otherwise use 0.8.  Or maybe support both
with a configure switch.  However, either of these would mean taking a
lot of developer time that would seem much better spent fixing a
single version.

It is somewhat unfortunate, but the lateness of the issue will
probably force some difficulty in the schedule.  We may need to shove
feature/module freeze effectively back a week (but not doing the same
for subsequent dates) or being lenient with GStreamer-related freeze
break requests at first to get this straightened out.

So, thoughts?  Opinions?  Ways out of this mess?



[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]