Hi Miller !
Looking through the code I realized that setting fragmentsizes is disabled in the Linux version. This is actually a good thing, as pd is more likely to run on any sound hardware/drive combination.
Still it would be good if there is a commandline option to tune the soundcard buffer size (prob through the -audiobuf flag).
This would make it possible for people to decrease latency.
As ost reported, he still has problems with audio on his notebook. This could be due to the GETOSPACE/GETISPACE calls in the calcspace function. (As I remember I had problems with this calls returning wrong values).
Is it possible in these cases to block, or maybe better use select ? Together with a "-very-stupid-soundriver" flag in the startup options.
This is just a guess about the remaining soundcard problems, but as ost said that his previous version of pd, which probably blocked in theses cases, worked for him it might be a valid one.
On the other hand, the right way would be to fix the driver :)
Guenter
PS: Congratulations to version 0.26, I really like it.