On Aug 7, 5:03 pm, Piet van Oostrum <[email protected]> wrote: > >>>>> Michael Mossey <[email protected]> (MM) wrote: > >MM> Ah yes, that explains it. Some of these long computations are done in > >MM> pure C, so I'm sure the GIL is not being released. > > Is that C code under your own control? Or at least the glue from Python > to C? In that case, and if the C code is not manipulating Python objects > or something in the Python interpreter, it could be changed to release > the GIL during the computation. That's also how Numpy does it, IIRC. > -- > Piet van Oostrum <[email protected]> > URL:http://pietvanoostrum.com[PGP 8DAE142BE17999C4] > Private email: [email protected]
I don't have any control over this code, and it's easier to solve my problem in other ways. I just put a sleep() call between calls to the C library, and that gives the network thread enough responsiveness for my particular task. I am grateful, anyway, to understand why this kind of thing happens. -- http://mail.python.org/mailman/listinfo/python-list
