fishsoup

Just got back home from day 1 of the GNOME Summit. So far, my overwhelming lesson is that if you put 50-odd GNOME hackers together in a couple rooms with power and network, they know exactly what they want to do. They want to a) hack and b) talk to each other. So, the bulk of the day was people doing exactly that. We had some good more-structured sessions, but they definitely weren’t the focus. I hope people will blog about what they were hacking on.

The venue: at the Royal Sonesta Hotel Boston has worked out well so far. The staff there really has the event thing down: when I got there this morning at 7:45, all the furniture, equipment, network, was set up exactly as ordered. Coffee appeared magically on schedule. And so forth. Beats the heck out of lugging a gigantic vat of Dunkin Donuts coffee in the back seat of my car, as I did a couple of years ago. Many thanks to Rosanna for getting this all arranged so well at the last moment. I do hope we can plan ahead a bit more next year and get a University venue as in past years … it’s certainly more cost effective, and also perhaps a more natural environment for hacking.

I attended three sessions during the day. My team from Red Hat had a session on the online desktop in the morning; we spent a lot of the time just demoing it. You never realize how much you’ve created until you stand up there and show it all. Then we had a good discussion about possible extensions and ecosystem issues. Thanks a lot to Asheesh from Creative Commons for keeping us honest and bringing up the question of how the online desktop relates to software freedom… it’s an important point. We can’t just think that a free shell to a bunch of closed web services is the end goal.

After that, there was a GNOME Love/Getting Started session that Thomas Hinkle suggested (Thomas is the author of the current Soduku in gnome-games, and hopefully much more in the future.) The concerns were pretty much what you would expect: the difficulty of getting things built, the lack of documentation about how everything fits together, patches languishing in bugzilla, and in general the difficulty of knowing how to go from having an itch to scratch to actually scratching it. One thing that was brought home to me personally was the value of the GNOME Love idea. I must admit that I’ve always secretly considered it just a bit wonky. What’s there to know? You just grab the sources, start coding, and if you hit a problem, you hop on #gtk+ and ask it. A view obviously highly colored by the fact that when I started out there were no experienced GNOME hackers. We were all puzzling out automake together. By the way, if you are in the Boston area, new to GNOME hacking, or just thinking about contributing you really should come down tomorrow or Monday. It’s free, there’s no registration, and a great way to get a feel for the community.

Then in the afternoon, I led a session on JHBuild, starting from some of my ideas in my last post. I don’t think anybody at the session other than me was a current user of JHBuild, so the discussion was a little theoretical. Which definitely drives home the need to revitalize building GNOME from the upstream sources as an activity in the GNOME community. The number one desire among the people there was making JHBuild better at building just part of the stack. Other things discussed: creating packages or vmware images from JHBuild results; installing JHBuild as a system package; build parallelization; building in a chroot or Mock environment to keep system dependencies predictable. David Zeuthen even suggested ‘jhbuild bisect’. One topic that got a lot of discussion was making it easy to try out a patch: given a patch in bugzilla, how do you make it easy for an end user to try out that patch, either by rebuilding their system package with the patch, or via JHBuild?

Talked to a lot of other people as well … talked to Travis Reitter and Robert McQueen about using online.gnome.org and the desktop data model to organize and merge together your contacts from the different Telepathy backends. Talked a bit to Jeffrey Stedfast about optimization of the use of Pango in Moonlight. And so forth. The fact that I was able to spend a lot the day in sessions and talking GNOME with people I think shows how smooth the setup has been. The only real issue was herding everybody out of the rooms at the end of the day. On the equipment list for next year: a megaphone and a sheep dog.

I wimped out and went home after the last session instead of sticking around and heading out to dinner, so I missed a lot more good discussion. Things start bright and early again tomorrow morning. The plan for the 9:00-10:00 slot is to have a “checkpoint” session to see how things are going. Come prepared with demos of cool things you’ve hacked up and suggestions for how we can make the next two days even better.