Re: Minutes meeting Oct 26

[Vincent Untz <vuntz gnome org>]

Le lundi 27 octobre 2008, à 16:48 +0100, Olav Vitters a écrit :
> + release team membership changes
>  - nobody was willing to leave
>  - fredp was acceptable, some discussion regarding 9 man team being too
>    big
>    @vuntz to talk to fredp

Good, will do this now!

> + looking back at the 2.24 cycle:
>  - if vuntz is not around, the release is in trouble
>   - bkor and mclasen mentioned too short timeframe between deadline and
>     release

Nod. I've been thinking the same. I don't think we can change the
deadline for tarballs (it's important to have the week-end for late
people, and a last week day for those who don't work the week-end).

It might be possible to change the release day, though: from Wednesday
to Thursday or Friday -- the week-end would be a bit late.

We'd need to see what is the main issue with the short timeframe:

 + building? Right now, it's an issue, but this looks fixable: that's
   something we could automate a bit with the build brigade or other
   stuff.

 + smoketesting? We actually know of the major bugs quickly thanks to
   the distros

 + reaction time for major bugs? That's probably the main problem as
   far as I can see: if a module has a major bug, we need to take a
   decision quickly and we need the maintainer to have time to do things
   quickly...

>  - gdm decision process was bad
>    @ by lucars: in future, if situation is not clear enough for too
>                 long, say no to the (rewritten) module

Nod. Another issue was that we had nobody in the release team feeling
that he "owned" this decision -- I thought it was Andre and Andre
thought it was me for a long time.

> + setting in stone the 2.25 schedule
>  - andre already mailed it
> 
> + new module proposals
>  - external dep: libproxy 
>    doesn't do the actual proxy handling, only finding out which proxy to
>    use
>  - gnome-user-share
>  - likely: notification daemon, NM

Oh, didn't know that. Interesting.

>  - ask to apply: (?)
>     - conduit
>       @andre will ask
>     - webkit-gtk
>       Epiphany isn't sure if webkit port will be ready in time.
>       Yelp+devhelp are likely ok.
>       Evolution is a lot of work, focus now is on de-bonofication. HTML
>       widget is used than just Composer.

I've followed the webkit-gtk a bit since Boston Summit. There were big
issues in how development went in the past few months, but people are
trying to fix it. Having things ready for 2.26 is a big challenge, and
we shouldn't rely on that. But it might happen, so we just need to be
ready in case it does :-)

>       @bkor to ask to concrete thoughts, from all modules
> 
> + tasks for 2.25/2.26
>   bkor proposed to get some doable non-forceful tasks that are promoted
>   for a specific release
>   - degnomeification (removal of libgnome*)
>   - if glade-3 with gtkbuilder, glade->gtkbuilder
>     Some doubt if glade-3 is ready in time. Also, glade convertion tool
>     (used by some modules) fails for Nautilus.
>   @mclasen to work on other good task proposals for 2.26

Awesome!

> + rewrites during 2.25/2.26
>  - existing rewrites should be finished before starting new ones
>  - mclasen mentioned that gconf might be rewitten (dconf?)
>    --> likely to occur for 2.28
>    @bkor to  ask desrt + codethink for plans/roadmap
>  - 'webkit' --> see new module proposals
>  - hal -> DeviceKit
>    --> likely to occur for 2.28
> 
> + release assignments
>  - bkor declined from all
>  @mclasen: 2.25.2, 2.25.4
>  @lucasr: 2.24.3, 2.25.5, likely: 2.25.91

Aaaah, thanks! Anybody else who wants to take a release? :-) I can try
to take the remaining ones, but I'd be happy to have even less releases!

> + GNOME 3.0 discussion

Sorry for this, especially since I'm the main blocker :/ Will try to
finish my stuff today & tomorrow.

> + AOB:
>  - not really part of meeting, but leio mentioned frustration from
>    distribution POV regarding rewrites
>     - gvfs being unstable during 2.22
>     - gvfs && gnome-vfs mix
>     - gnome-session not saving the session
>     - loss of functionality after rewrite

FWIW, I also got similar complains from PLD.

>  - also from leio: external dependencies being optional
>     - explained that external deps are ok to be mandatory
>     - leio gave example of vinagre depending on avahi for feature which
>       not everyone wants
>     - outcome: compile time things add to maintainers burden

Not sure what's the issue here? It's quite clear that it's fine to
depend on external deps in a mandatory way, isn't it?

Vincent

-- 
Les gens heureux ne sont pas pressés.