On Sun, 16 Dec 2001, Chris Kennedy wrote:

> In the latest CVS, if I put the sound_server to win32p or to sdl...it plays
> a little bit of music and then crashes. (Attempted to access memory at <hex
> number here>, & the memory could not be written) Another null pointer?

I'm not seeing this right now.. maybe this was fixed by Alex's checkin?

> I reverted back to an earlier build from a few days ago to test with the
> fresh install. The music played and Reverb worked on the intro, but it held
> the notes all the way into the scene where the door opens & continued to
> hold the notes until Roger hits the ground. (snare hit) In one instance, the
> music even continued to play into the next song. (which was really weird)
>
> Also - throughout all the versions (using win32p - the default) the music
> has been inconsistant upon each instance of freesci. Let me explain.

This has been a constant problem, and I had corrected many of the
potential sources for this problem. I still have no idea what the problem
would be, unfortunately. It seems like a race condition, but I've never
been able to track it down to fix it. If memory serves, the initial
soundserver commands (or data) were being sent before the soundserver
before the soundserver had initialized the queues. I was trying to figure
out how to make this happen in the right order and never figured out a
proper solution.

Any fresh ideas on this would be appreciated. This whole problem was the
inspiration for Alex to do the event based soundserver, which appears to
have it's own set of problems at the moment. These (and the PQ2 bugs) are
what I'm looking into now.


--
http://www.clock.org/~matt


Reply via email to