Time for a third “Debian XSF News” issue!

Debian XSF News #3

• Julien Viard de Galbert keeps on triaging bugs: TXBW9+10+11.

• After having taken some advice from Sven Joachim, I uploaded mesa 7.10 to experimental. I made both dh_install and dh_makeshlibs stricter, which triggered a FTBFS on i386 and sparc respectively. The former is fixed in git already, the latter still needs to be fixed. mesa 7.9 was already failing on non-Linux ports, and that’s still happening for 7.10.

• I uploaded xkeyboard-config 2.1 to experimental. It is the source package for both xkb-data and xkb-data-udeb, which contain XKB configuration data. As mentioned in the changelog that’s a big jump (3 upstream releases at once). While preparing it, I noticed a significant size increase: the udeb grew from 3.1MB to 3.5MB. Since size matters in the Debian Installer, I tried “compressing” the files shipped in the udeb, getting rid of a bunch of extra whitespaces and comments. A few tests seem to show no regressions, and the size went down to 2.6MB. Yay!.

• I did some heavy bug triaging myself, attacking some source packages: xkeyboard-config, mesa, and xorg-server. All in all, I managed to send 80+ mails in less than 16 hours, meaning 5+ answers per hour in average. I spent some more time after that, reaching 150+ mails in less than a week. Many bugs got closed, which can be seen on Mike Hommey’s BTS graphs. The most interesting graph can be seen below.

• I closed a very old bug.

• I uploaded a new release candidate for xorg-server 1.9.4 to experimental. Nothing fancy on the upstream side (mostly XQuartz fixes), but on the Debian side, an update of the bug script. As a reminder, that’s the script run by reportbug to collect information about the system. That should help us spot more common issues.

• The bug script update was cherry-picked into xorg-server 2:1.7.7-12 for unstable, along with two bugfixes. If everything goes fine, we might try and get it accepted for squeeze’s r1 (in other words: the first point release for squeeze).

• I finally managed to get mesa to work on my Intel board enough to let me test wayland. Having a look at the architecture (it has pictures!) and at the FAQ should help interested folks to understand what this is all about. I opened an ITP for it, as well as one for libxkbcommon. The latter is already uploaded and sitting in NEW. We’ll also need GL support in cairo, so I requested it.

• I should stress a few points: wayland is really at an early stage of development; all that is going to be shipped for now is a sample compositor implementation and a few sample clients for people to play with. Maybe wayland is going to replace X some day, but it should be noted that it comes with heavy requirements, like GL/EGL support. It’s nice since many features and bugfixes are going to land in mesa, but that means that it’s going to be available on a restricted class of hardware only, as opposed to X’s ability to run more or less anywhere. This also means that early testers might need a pretty recent mesa snapshot. I’ll try to make that easier, or at least well-documented. (Many thanks to Darxus for the review.)

It’s been a year

On a personal note, it’s been a year to the day since I first looked into X. After having hacked on the Debian Installer to make it use X.Org instead of DirectFB, I did some heavy bug triaging, resulting in a drop in the xorg-server bug count in March. The same happened past week as written above, resulting in a second drop. In the meanwhile, the bug count remained more or less stable, since we try to reply quickly to new bugs, and since Julien Viard de Galbert does bug triaging on a weekly basis:

One might ask: what does “maintaining X” mean?

• Packaging mostly everything X-related.
• Taking care of incoming bug reports as well as older ones.
• Working with upstream (which is going to be the subject of a later blog post), as well as coordinating with Ubuntu.
• Keeping an eye on the Graphical Installer.
• Writing documentation targeted at both users and maintainers.
• Advertising XSF’s work.