[fluid-dev] Some more thread safety commits and stuff

[josh]

I just committed a bunch of multi-thread safety changes to the  
fluidsynth_event_queue branch.  I tested things briefly and all seemed  
well, but there could still be some bugs.

Notable changes:

fluid_hash.[ch] re-written from recent glib and now has the full power  
of a generic hash data type (not hard coded to string keys and  
fluid_settings_t values as before).

Added SoundFont reference counting, presets now add references to  
SoundFonts.  SoundFonts are only unloaded after all presets relinquish  
their references.

fluid_settings_t is now thread safe.

Converted fluid_mutex_t to use glib's recursive mutex, to prevent  
recursive deadlocks (possible with fluid_settings_t update()  
callbacks, if any settings functions are accessed in the update  
handler).



I realize now that we have probably missed the Ubuntu deadline that  
David had mentioned.  In hindsight, we should probably have released a  
1.0.10 without the more ambitious thread safety changes.  I'm going to  
continue plugging away at making FluidSynth thread safe, which I think  
we are close to now, but we should perhaps still consider the option  
of releasing a 1.0.10 version.  When the thread safe changes are  
complete, there is likely a bit of testing that is in order to make  
sure there aren't any issues, since the changes are not so trivial.

As to the previous discussion about glib dependency.  I have once  
again flip-flopped on my decision.  It seems to me that the features  
we want to use from glib are minimal (hash tables, lists and threading  
related data types).  I started disliking the idea of using glib  
directly, since FluidSynth is already designed as a library that is  
supposed to handle out of memory conditions.  I don't like the idea of  
another application being terminated by glib, when a memory allocation  
fails.  So I would like to port the needed code from glib to  
FluidSynth and change as necessary.  This will also resolve any build  
issues or other pains caused by having to build glib and make the  
embedded FluidSynth case easier.

Testing results on the fluidsynth_event_queue branch would be  
appreciated!  Perhaps we should merge it back into trunk soon?

Comments?

Best regards,
Josh