GUADEC 2010: "Building a strong post 3.0 Gnome story"

This is the talk that was hijacked by the release team to make the announcement about the delay in releasing Gnome 3.

Gnome 3: Where we are

Vincent Untz began by reminding us that there is a lot of excitement around Gnome 3 and Gnome Shell. To encourage the initial development effort around Gnome 3, the release team boldly stated 2 years ago that Gnome 2.30 would be Gnome 3. The community then reached a level of activity never seen in years. It was a very aggressive schedule, and it was meant that way: perhaps the work would not have started or not have been so intense without it.

However, Gnome sticks to a predictable 6 months schedule, and quality matters more than anything else. We want people to love Gnome 3. After waiting 8 years since Gnome 2.0, we want this release to be amazing. That is why the release team decided to move Gnome 3.0 release date to March 2011, having closely following mailing lists and using GUADEC as an opportunity to get the latest from every team.

  • Many modules, such as Gnome-shell are nearly there and could be okay for a release in September. But we want this release to be amazing, not simply okay.
  • Platform-wise, most people have stopped using the deprecated stuff, which is good.
  • Some modules will need more time to migrate, such as gsettings and the Human Interface Guidelines update which would not be ready in 2 months (Note: I attended a lightning talk about this and I like what they're planning on doing with this: a series of design patterns rather than a gigantic monolithic document that is hard to parse. They only started this the week before GUADEC though!)

The work for September

Some may worry that this is a bit late for such an announcement, and too much work to prepare Gnome 2.32 in September. Some modules will need to switch back to Gtk2.

There will also be a Gnome 3.0 beta release in September (\o/). The release team wants everyone to start using it, play with it, and report bugs. Application developers should start porting their application to Gtk3, as it will take time to get there and it would be better to start this work in September rather than next March.

The talk ended with a list of TODOs for the release team.

  • Revert to gtk2. They believe it was not that hard to move to 3, therefore it should not be hard to move back to 2. They're happy to help if people are not sure how to do this. Ideally this should be done though a new flag, to give an option to compile with gtk2 or gtk3 so that people focusing on the gtk3 effort may keep working on it.
  • They will keep pushing for Gnome 3, Gnome 2.32 is only a step. Keep porting!
  • Some modules such as Shell will be in feature freeze, to polish and finish what they've started.
  • The release team will also push the community to implement designs. Good mockups are being made and we're happy about it, but then nothing is done about them.

The questions part of the talk was interesting albeit a bit tense.

Gnome 2.32

The first question was about the 6 months release schedule, and whether it really was a valid reason to have a 2.32. What would be the content, wouldn't it only take time away from people's work on Gnome 3? Maintainers cannot be forced to do anything, and if they have better reasons to work on 3 it's okay. Perhaps other maintainers will have built new features and will want them to be used by people without having to wait 9 months. The audience still worried about the maintenance burden and cost for people already depending on Gtk3 features, but the speaker seemed confident that switching back to 2 was mostly about changing some configure options, and that no one had to do if it was too much overhead.

Still maintainers in the audience insisted that having both 2.32 and a Beta meant twice the work. Why spend time on something that matters less and less? They also disagreed that having gnome-shell and 2.32 in the background was a good idea, as it's not how the design is done nor how it's meant to work. Gnome 3 should be more than the Shell. Untz seems to believe that many features do not rely on Gtk3 or the Shell.

Maintainers contended that it wasn't the same amount of work for an application like Cheese compared to the core desktop. There is a burden to support 2 desktop environments in parallel. To which Untz answered that feature freeze was still planned for the following week, leading to a "what's the point?" question. Once again, if people don't want to do it, they don't have to.


Concerns about the marketing message were quickly relieved by someone from the marketing team, who said that the "Beta" label generates excitement. Anyone with a feature that is stable enough should stick it in, although the control centre is another matter. We still want to ship a quality product, if it only has half the features it's okay as long as they're stable. It will generate the buzz.

There was a question about working with distros to help ship 3.0 in a way that is easier to install, to encourage people to try out the beta. There are instructions for many distros already, but it was suggested to also offer clear and prominent instructions on the Gnome front page once the beta's out.

The last question was about what the schedule would look like for the next 6 months. Although the schedule will be mostly normal, with new modules and new features, some rules will be similar to now in that new applications won't be accepted in just because they're great. Likewise, it doesn't make sense to have a freeze for everything, but for some modules it does.

Final words

Keep working on Gnome 3.0!