On Tue, 22 May 2001, Taylor Barratt wrote:
> > did you try the last update? I put in a Sleep for the flush function which
> > may help the SysEx problems you described. Please let me know if this
> > works any better for you before I begin the plug-in/module conversion
> > process on win32.
> Sorry 'bout not trying it earlier, I had a long race weekend. Anyhow I
> tried it out last night and sure enough I don't have any problems with
> loading sysex patches anymore (woohoo!). So yet another step is overcome.
> *cheer* Can't wait for the speed problem to disappear.
I've observes some additional behaviour using perfmon on win2k that
verifies what I observed on win98:
Sometimes the "main" thread is thread #4, that is, it appears to be
started after the soundserver thread. When the soundserver thread is
started first, music plays fine from the start.
When the main thread is thread #1 and the sound thread is thread #4,
music is "slow". It's quite possible that is misses the first few events
in this scenario, causing the problem.
My guess is that this is a timing problem with SDL threads on win32.
Does anyone have an idea where I should start looking to fix this problem?
Insert some sleeps? Make the sound thread initialization happen sooner?
--
http://www.clock.org/~matt
