Hacker News Reader

And for further information: http://jessenoller.com/2009/02/01/python-threads-and-the-glo...

Ultimately, while the implementation of the GIL is inefficient, and does harm you in the case of processor-bound computation, they're still useful in many cases where you're dealing with I/O (file, and socket).

Even as the multiprocessing maintainer, I'd like to see a more robust threading model within python - unladen swallow (http://code.google.com/p/unladen-swallow/) may be able to offer us a world without a GIL soon(ish), but even barring that, getting some work done on improving what we have today would be nice.

"""•Bias : For parallel computing involving heavy CPU processing, I much prefer message passing and cooperating processes to thread programming"""

Of course you prefer it .. you have no choice!

Baloney. Many of us who do want better threading in python still prefer message passing and processes. That's nothing new. Different tools work well for different jobs.

Message passing and cooperating processes do have a lot of real advantages: - race conditions are almost eliminated - deadlock is more easily avoided - no need for locks

Also on blip.tv: http://blip.tv/file/2232410

Inside the Python GIL