Le samedi 25 août 2012 11:38:47, Vincent Pelletier a écrit :
> Any idea of ways to debug this problem further ?

Trying with pypy ("just to see"), I got even more reproductible segfaults - 
even with valgrind.

Turns out, I was not keeping strong references to ctypes buffers, which get 
very quickly collected with pypy and memory reused for something else, leading 
to a quick crash.

Now, I don't get yet why cpython was not crashing more often, even if I forced 
gc.collect() . Less memory recycling maybe ?

Regards,
-- 
Vincent Pelletier
-- 
http://mail.python.org/mailman/listinfo/python-list

Reply via email to