> ftfind() returns a a pointer to a function [a FUNC struct] for a given
> table # - so it looks like you would be able to access the function data
> and read/write/mangle or whatever.
>
> this of course could be dangerous, but if the host app knew what it was
> doing it should be ok i think [as long as table lengths were not changed]

Yup that's exactly what I'd like!

> ive been planning for a long time to get graphic editing of tables in
> MacCsound, and was just planning on continuously sending f statements,
> your idea would be smoother - then again if you are changing breakpoints
> for a gen5 or something, you would probably need to recalculate the
> entire function --  i do this in one of my demo csound~ patches, and it
> actually works in realtime fine -- obviously if you are oscillating that
> function to produce audio or something, you usually get a click, i cant
> think of an easy way to 'lowpass' this for all gen types to get rid of
> the clicks...

Yeah, it could be misused for sure, but like you said, if the host app knew
what was going on it could be done properly. For example I had planned to be
doing some writing to the tables and this would be timed properly so as to
avoid havoc. ( The tables aren't holding audio data ). Plus, after the data
is read out from them in csound it can be interpolated.

I'd love to hear of any progress on this front.

Iain