Re: [fluid-dev] sequencer / synth bug ?

[Antoine Schmitt]

Hi Miguel,

yes I had seen your answer of course, thanks. Without meaning to minimize it ;-), I thought that mentionning a design problem would generate more traffic on the list.. But maybe it just means that everyone agrees with the analysis.

Also, remained the issue about the 1ms timer not being really a 1ms timer (on Windows).

About your solution, I understand that it is a large overhaul of the fluidsynth engine, where, in addition ot other changes like going to C++, all calls go though the sequencer, and the sequencer is linked to the audio clock.

What I propose is simpler, and, if I am not mistaking, would be a simple variation of the existing fluid engine. I could even be considered like a fix.

Do you (or anyone else) see a problem with the solution that I propose ? If not, I'll try to implement it...

Thanks a lot,

Le 19 juin 07 à 13:23, Miguel Lobo a écrit :

Sorry if I'm being obtuse, but did you see my reply to your original post?

http://lists.gnu.org/archive/html/fluid-dev/2007-06/msg00001.html

In short, I completely agree with your diagnosis and the solution I'm using in my project (derived from FluidSynth) is essentially the same as yours: using a single clock derived from the audio output.  The only difference is that in my design the audio clock is not necessarily real time.

Regards,
Miguel

++ as