Comment 19 for bug 999111

Sorry about the size of the diff, this is my fault. I just kept going on and on, merging all of trunks bug fixes upstream without much regard to the SRU process. Some of those changes, like fixes for the glitchy opacity transitions touched a lot of code, but generally simplified things and are supposed to be more robust as a result. Similarly, last minute changes of Precise allowed us to rip out clunky workarounds, namely detecting iconifying by libwnck instead of formerly broken GTK3.

I understand your frustration, perhaps 0.97.1 really isn't SRU material. It was well intentioned from our side, though. Regression testing for all releases is (still) done manually by checklist in four desktop environments: unity, unity-2d, gnome-shell and gnome classic (with and without compositing). If it helps in any way I'd send you the list.

Going through all the bugs again and creating test cases would take a lot of time. Most bugs and their fixes are only briefly documented in bazaar. If the SRU depends on a complete set of test cases we'd probably have to forgo it. Onboard 0.97.0 has been rather resilient anyway and people can get the update from our PPA(s) if needed.

I think at least one good reason for a SRU may be the ongoing troubles with translations crashing Onboard (for various reasons). See fresh bug #1013279 for example, but there have been plenty of others. The proposed update is much more robust in that regard and simply falls back to English for unexpected translations. There are usually workarounds available for the current version too, though.