It has become a tradition now: Google is organizing a Summer of Code this year again, and quite a big number of organizations will be involved in mentoring students. We can certainly expect some good stuff to happen thanks to this program! Some students have already started applying: the application period began last week and will end next Friday. Oh, by the way, dear students: submitting your application a few days before the deadline is a good way to make sure that potential mentors can love you ;-)

Of course, GNOME is participating as a mentoring organization, as well as openSUSE. There are some quite good ideas in both cases, but of course, students are welcome to apply with their own idea. When applying, keep in mind that one of the most important things is that the application is well-written and detailed, and comes with a good plan. Also, in the case of GNOME, we ask students to provide a patch -- this is a step that lets mentors easily know if the student can at least get the code and compile it ;-) (and the quality of a patch, or the way it's written, actually often tells us a lot about the student)

A big difference for me this year is that I'm not an administrator on the GNOME side (I agreed to help a bit for openSUSE, but Zonker is doing all the work). Last year, Adam and Sandy did a great job for the organization, so it was not a big surprise to see them step up for the task this year!

Of course, even if I'm not an admin, I'm still curious; so I'm looking at the student applications and adding a comment here or there, but nothing really big. I might mentor a project, though, if the idea I proposed for openSUSE seduces a student and gets well-ranked, or if a student comes with something I'd really love to see and where I could help! I guess we'll know in a few days ;-)

Lucas sent the announcement earlier today: the GNOME project will move to git.

This announcement highlights the recent steps that lead to this decision: the DVCS survery and the analysis of its results. Of course, all this was discussed on mailing lists (okay, it quickly went from discussion to flamewar ;-)). It was not a decision that was taken after onl a few thoughts, and I guess it's good to remind people that using a DVCS is something that was in people's mind since quite some time. For example:

• gvfs development was first done in git, back in 2006.
• people suggested to have git as an option back when we were doing the subversion migration, and this was discussed in various places since then.
• we had a git mirror since last year (alongside the previously created bzr mirror).
• it's also worth pointing out that the GTK+ team considered git in the past, or that NetworkManager moved away from the GNOME infrastructure specifically to use git while other projects like empathy are using git via git-svn since quite some time.

Note that I mainly put git-related things in the above list; I'm pretty sure you can find some bzr/hg/etc. items that would also show the community interest in a DVCS. It turns out the chosen DVCS was git and it's simply because the majority in our community prefers git. Some people might wonder why we're not letting developers choose another version control system if they prefer another one; the answer to this is easy: to keep our infrastructure maintainable, we choose to only have one version control system. It doesn't mean other systems are bad. Really.

We decided to make the migration happen in one month, after 2.26.1. The original idea was to make the switch after 2.26.0, but we wanted to give some more time for people to double-check the great work that the git transition team has done. Furthermore, the month between .0 and .1 releases generally sees quite some activity to fix bugs filed after the .0 release and for translations, so it made sense to wait for 2.26.1 to be out before breaking the habits of quite a few people ;-)