On Sunday 11 January 2004 10:56, Kjetil Svalastog Matheussen wrote:
When using realtime priority on the audio thread, any non-realtime thread shouldn't cause a dropout... At least in linux. Perhaps it can in macosX?
But it doesn't look like you're using a seperate thread at all (?) I'm using pd-0.36-devel still, so maybe things are different now, but if so that would have been quite a radical reworking of PD, to have the message processing be in a seperate thread than the audio.
So I'm not too eager to add threads-support. However, guile does support threads (documented in the manual), so perhaps the pd-* functions should be made thread-safe? If there is a demand, it shouldn't be to hard to do.
Again, I'm a bit confused. I had always thought, that unless you launch a thread on your own, that PD was single-threaded, and that it relied on message processing to be fast enough to not cause audio dropouts - hence there seems to be no reason at the moment to make your functions thread-safe. Now of course, it would be nice if your external worked off-line in an other thread, in which case re-entrancy might be then a factor.
But I think that running this external in a second PD connected to the primary one with netsend and netreceive should work, right?
Should work.
Larry