On Mon, 31 Dec 2001, Alexander R Angas wrote:

> That's interesting. On my machine, it plays slightly faster than before, but
> still too slow.

It's still slightly too fast or too slow, I think. I'm doing everything
right, I think there's some latency involved using PostMessage or
something somewhere else..


> > Whoa! This used to drag the songs on, but now it belts them out at super
> > speed. I know changes were made to the timing/sleep routines. Interesting
> > things happen on playback. I will use SQ3 (the one I have been
> referencing)
> > to explain playback timing.

I have been using SQ3 for my testing, please try again with latest CVS.
I believe it sounds better, but is still not perfect. Fortunately, I
think it is on par with the polled soundserver at this point since it is
at least consistent.

> > Once Roger gets out of the door, the garbage scow theme plays ultra fast.
> > After it gets to be about 5 seconds in, the music just drops.

This shouldn't happen anymore, and I can't reproduce it with my latest
changes. Please let me know if you still hear this happening.


> > Here is some stuff from my console window:

> > timeEndPeriod(1) failed in sci_gettime

Fixed.


> > Warning: Attempt to restore invalid handle 2800
> > Warning: Attempt to restore box with zero handle

This is not me, but I see this as well.


> > GFX-SDL 1091:Received unhandled SDL event 0011

I finally fixed this.


> > Of course, it still slows down on a large amount of notes. I am running a
> > Pentium III cu at 700 Mhz, and I know it was pointed out earlier that the
> > event server was processor taxing.

I've noticed this as well, I can only assume it's because more messages
are being sent to the invisible window we use. Maybe we should eliminate
sending messages this way?

I was thinking in the callback function we could just call
sci_get_from_queue() instead of posting messagses to the invisible
window.. comments/suggestions?

Christoph: please review my changes and let me know if they are safe to
check into the 0.3.3 branch.

Thanks for testing!



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


Reply via email to