FT: Intro sound regression
The Full Throttle intro doesn't work as well as it used to in recent CVS snapshots. There are numerous drop-outs in the sound, and at least one sound (I only watched the beginning of it) repeats.
I think it may have started with revision 1.14 of smush_mixer.cpp, where the buffer size was decreased from 2,000,000 to 100,000. If I turn on debugging output I get lots of "mixer full" messages. The messages don't go away completely even if I revert the patch, though I no longer hear any sound errors.
But I don't *want* to revert the patch, because I agree with the reasoning behind it. I just don't know why it doesn't work in reality.
Ticket imported from: #765800. Ticket imported from: bugs/938.