MOTU Application for Fabien Tassin (fta)

[Fabien Tassin]

Stephan,

[ sorry for the late answer, you replied only to motu-council and I'm not
  a subscriber of that list ]

According to Stefan Potyra:
> 
> during last FeatureFreeze, I recall that you were a backup for Alexander 
> acknowledging/rejecting excpetions, or that at least I asked you once or twice 
> for guidance.

I remember.
 
> Since I wasn't able to track any freeze exception requests in regards to teams 
> to which we delegated responsibility, I'm interested to know how good this 
> worked for the mozilla team. Do you have an estimate how many exceptions were 
> granted for the mozilla team? 
> Do you believe that the model (i.e. delegating freeze exception requests) 
> worked well? 
> Did you find issues with that model, or issues in regards to freeze exceptions 
> that you think should be addressed when intrepid feature freeze starts?

I don't have a precise number of freeze exception requests we got for hardy
but that was not high. I remember some xul-1.9 transitions and some extensions.
But maybe Alexander got more on his side. I think it worked well and I heard
no complains. IMHO, it makes sense that freeze exceptions are delegated to
specialized subteams, as they (should) know the specifics better.
 
> Finally, as I asked the same question to Dustin in regards to the server team: 
> Do you think the mozilla team's integration with motu in general is working 
> well? Where do you think that the integration/cooperation could still be 
> improved?

A bit of background first.

The Mozilla team is quite small and we are specialized (some of us do bugs,
some do packaging). We also have a new team for extensions, but it's still
young and need some work.

We have packages both in main (firefox*, xulrunner-1.9, tb..) and in universe
(seamonkey, sunbird, prism, tons of extensions..). Some are low maintenance
(either because they are maintained by Debian, or because they are small
or staled projects), some require a lot of efforts.
We almost never use REVU, we work in/with Bazaar branches instead.
The approach is IMHO different, REVU tends to be a way to please lintian and
reviewers for the package as a whole, while with branch reviews, we focus
more on incremental improvements and bug fixes, each as unitary commit.
At least, that's my view of it.

We have multiple branches per product based on the distribution and maturity.
We tend to prefer grouped changes instead of pushing releases for every
single small change. Bazaar is very helpful to make our work visible
either through LP integration, or by using links in bugs.
Last year, after a lot of improvements in some packages, I started a project
to move the common parts of Mozilla packages to a unique dev package.
It's now called mozilla-devscripts and it hosts several scripts such as a
tarball creator, an install checker, an addon packager and a lang-pack
exporter. This is supposed to make our packages easier to maintain, yet it may
look like more dark magic to some.

We sometimes receive contributions from the community as debdiffs posted
in bugs. I don't remember anyone outside of the team proposing branch
merges or reviews (with the noticable exception of extensions).
I would love to see MOTUs contribute here, especially when we are lagging
behind. A granted merge could serve as an ACK.

Almost all of our packages are sponsored by Alexander, that adds load on him,
which is not good as he could spend his valuable time on something else.
Even if we go through main/universe sponsors, it almost always ends up
on him anyway. The detour usually adds delay, sometimes 2 or 3 weeks, which
could be frustrating for contributors. This loop doesn't really make sense.
My point is that packages with a specific home should be handled by the
corresponding team, knowing that contributions are always welcome.

My feeling is that Mozilla will be more important in Universe in the future.
I packaged some new stuff, such as Flock, Songbird, Firefox 3.1 and its
XUL 1.9.1 SDK, and I really would like those to enter the repository.
Alexander is well aware of the progress of those packages and we agree on
what shape those packages should have before they are considered ready.
All in one, a good part of our packages is indeed already or destined to be
in Universe.

I hope I addressed your questions and maybe clarified a few things.
Feel free to ask if you need more information.

/Fabien