Elijah Newren wrote:
An alternate strategy which is actually making [bug prioritization based on frequency of occurrence] possible is the cooperative bug isolation project (http://www.cs.berkeley.edu/~liblit/sampler/). Check it out, it's really cool and it has found and fixed some "random crashes" already.
Thanks for the endorsement, Elijah. :-) We have had several successes knocking off "random crashes" in Rhythmbox:
http://bugzilla.gnome.org/show_bug.cgi?id=130788 http://bugzilla.gnome.org/show_bug.cgi?id=137460 http://bugzilla.gnome.org/show_bug.cgi?id=137834Because of the way the system is designed, bug triage based on frequency of occurrence is implicit. We learn the most, most quickly, about the bugs that happen most often. That can be handy in a world where there are always more bugs than engineers.
What we really need now is more users, and more exposure to get those users. I'd be eager to discuss possible collaborations with anyone who thinks they can muster a few hundred or a few thousand users. I get data for my dissertation; you get higher quality software. Win win.