[fluid-dev] Voice-stealing logic, and more

[David Henningsson]

Hi,

I'm currently doing heavy work (more than anticipated) refactoring and
building up the new voice-render / midi state machine separation
infrastructure. For that reason, trunk is a little broken currently, let
me know if this causes problems for you (and I'll fix it up and create a
branch instead).

One of my design patterns is to split some of the behemoth objects into
smaller objects, in order to improve cohesion and readability. That way,
should I have thought something wrong, we are more likely to be able to
reuse at least some parts. In addition, the thread boundary logic is
quite straight forward. (Thanks for the ringbuffer implementation
Elimar, it came in very handy!)

Anyway, one of the things that are broken is the voice-stealing logic (i
e what to do if FluidSynth runs out of polyphony). I know there have
been thoughts about improving that anyway, so what are your thoughts in
that area - i e what requirements are there for that improvement?

// David