Lots of commit and bug tracker activity this month. As we're heading for the 0.2 release I spend most time doing bugfixing. Berzerka discovered that song-saving was flawed. Editing the patterns did not always took effect. Removing machines had leaks and ref-count problems. There were some long-standing gtk-warnings. All fixed in CVS. Finally that loop-playing problem got on my nerves. I was trying to fix that several times in the past already. Finally I found that it was a small issue with GStreamers adder element. Many thanks also to Henri Valta, who contributed several patches. All the new testing also lead to some new and some re-enabled unit tests.

Beside fixing, I was also working a bit on the features. The interaction controller now works for input devices (via evdev). I've tested it with a USB joystick and the harddisk in my laptop (via hdaps). Control devices are detected via HAL and shown in the menus. One can assign those to effect parameters. Using the accelerometer in the harddisk, one can control e.g. filter-cut of and resonance by tilting the laptop - how useless. There is one flaw to mention - the bindings aren't yet persistent, close the window and there gone.

Finally in gst-buzztard we have a first working version of a fluidsynth GStreamer source. Josh Green the author of swami made the initial port.

I will now start to prepare the release. Should be out in the coming weeks.

gstreamer

We got some data mutation suites for media files at nokia. We're currently test and fix gstreamer elements using them. This is really great, there are quite a lot of issues that would have been hard to discover otherwise. Unfortunately sometimes I can't fix issues. Since a while gst-plugins-bad has two midi-renderers - timidity and wildmidi. They wrap the libraries of the same name. Both libs are easy to crash and both projects seem to be dead :/ No replies to mails and patches are sitting uncommented in the tracker.

I also spend some time with working on the various build configurations switches for gstreamer. Most of them work again. Two issues left. First is generating bindings accordingly. I wrote some m4 macros for this, now they need to be used. Second issue is that building docs would fail if subsets are disabled. I'll ignore that for now :)