Dec. 9, 2011
3:18 p.m.
Le 09/12/2011 09:31, Robert Kern a écrit :
We have some global state that we need to keep, and this gets interfered with in a multiple interpreter environment. I recently got interested in multiprocessing computation with numpy and now I get scare by your statement ! Please don't tell me it is unsafe to launch multiple jobs (for instance with multiprocressing's Pool.map) just doing some ndarray arithmetic !
That's why I'd like to understand better the issue raised by Yang. For instance, what does exactly "multiple CPython interpreters" stands for ? Best, Pierre