[LAU] yoshimi records

[cal]

... maybe. I've gotten Record to the point where it hasn't done anything violent
to me for a while, but it still has plenty of potential for unexpected
shortcomings. The story so far ... 

- new Audio/Midi settings - default recordings directory and a checkbox for 32 bit
float wavs. The default is 16 bit shorts. The wavs are courtesy of libsndfile, so
that's a new dependency.
- Select your wav file via the 'Wav file' button. Opening an existing wav starts
in append mode.
- Orange rewind button does a rewind/overwrite.
- Green starts recording.
- Blue pauses recording.
Yes, that's a weird choice of colours. Send me some hex colour numbers and I'll
gladly change them.

To be honest, I'm struggling to hold focus well enough and long enough to give it
the sort of testing it needs. There's more than a few different usage paths and
despite best intentions, serious disciplined testing just isn't happening. So I'm
asking for some help with testing.

Background ... if record is running, at the point where audio is generated it
gets fed off to a fifo. The buffer on that is currently set to 256k. For no
particular reason I can think of I felt more comfortable with it at 512k. I've no
idea how low that can go before the mechanism chokes and splutters. Yet another
thread reads from the fifo and writes it to wav file. Does it properly manage the
integrity of wav files in all possible use cases? I think so, but I'm not at all
sure. How busy can the system get before audio and/or recording quality fails?
Don't know.

When you find it broken, the best I can promise is that I will try to fix it :-).
<http://www.graggrag.com/yoshimi/yoshimi-0.037.tar.bz2>

cheers folks.
Cal