Nick Coghlan wrote:
Paul Rubin wrote:

That's weird.  Preemption should happen every few dozen milliseconds
unless you've purposely increased the preemption delay.


To me, it smells like a call into a C extension which isn't releasing the GIL before starting a time-consuming operation.

Sorry, I think I misled you both -- "a long time" is actually the few dozen milliseconds of a normal preemption, which is only long relative to the single millisecond that the consumer typically takes to handle a request.


Cheers,

Evan @ 4-am

--
http://mail.python.org/mailman/listinfo/python-list

Reply via email to