Occasional segmentation fault when non-blocking function is interrupted by end of script about python-sounddevice HOT 6 CLOSED

I was trying to look at this but I quickly realized that sounddevice isn't actually a compiled extension which makes it more difficult to debug C issues like this. Is it much more troublesome to make it into a compiled extension?

from python-sounddevice.

I'm not sure if I understand ...

It would be reasonably simple to switch from CFFI's ABI mode to API mode, see http://cffi.readthedocs.io/en/latest/cdef.html.
This would change hardly anything in the implementation, most notably, the callback would still be implemented in Python.

If by "compiled extension" you mean manually writing C code using the Python C API, this would be an insane amount of work. Also, PyAudio already does that: https://people.csail.mit.edu/hubert/pyaudio/.
In this case, the callback would still be written in Python.

If you mean to keep CFFI (in API mode) but implement the callback in C, that's what I'm doing in https://github.com/mgeier/python-rtmixer/.

Finally, if you are talking about using the Python C API and implementing the callback in C, that's what pyo does: https://github.com/belangeo/pyo.

from python-sounddevice.

Sorry about the vagueness, I actually just meant using CFFI in API mode. All this would do is produce some debuggable C code.

from python-sounddevice.

OK. It should be straightforward, but I haven't actually tried it.

Do you want me to make a branch for API mode?

from python-sounddevice.

from python-sounddevice.

Fixed by #75.

from python-sounddevice.