I've merged a branch 'scanning-speedup' into next that doubles the speed of adding messages to the index. There was a lot of time wasted in parsing the email headers, which I fixed, and I tweaked a couple other things. Various other operations that load data from the message source, like viewing a message thread, should now be slightly faster.
I'm able to index around 70 messages/s on my home computer, which is still slow, but hey, twice as fast as 35. Time is split about 50/50 between scanning/parsing and Ferret index writing. Sadly, Ruby has a GIL, so I don't think threading these two operations will buy very much. (Though there might be complicated ways of getting around that.) -- William <wmorgan-...@masanjin.net> _______________________________________________ sup-talk mailing list sup-talk@rubyforge.org http://rubyforge.org/mailman/listinfo/sup-talk
