[Python-ideas] An alternate approach to async IO

Greg Ewing greg.ewing at canterbury.ac.nz
Wed Nov 28 22:28:29 CET 2012

Sturla Molden wrote:
>>Then why not just have one thread?
> Because of the way IOCPs work on Windows: A pool of threads is waiting on the
> i/o completion port, one thread from the pool is woken up on i/o completion.

But does it *have* to be used that way? The OP claimed that Twisted was using
a single thread that explicitly polls the IOCP, instead of an OS-managed
thread pool.

If that's possible, the question is then whether the extra complexity of
having I/O threads wake up the Python thread gains you anything, given that
the Python thread will be doing the bulk of the work.


More information about the Python-ideas mailing list