This issue tracker has been migrated to GitHub, and is currently read-only.
For more information, see the GitHub FAQs in the Python's Developer Guide.

Author giampaolo.rodola
Recipients Ikinoki, giampaolo.rodola, josiahcarlson, lasizoillo
Date 2010-04-19.19:55:11
SpamBayes Score 7.344205e-05
Marked as misclassified No
Message-id <1271706913.68.0.421411127812.issue6692@psf.upfronthosting.co.za>
In-reply-to
Content
The patch in attachment implements support for epoll() and kqueue() by adding a new "poller" argument to asyncore.loop().

However, I had a chat with Jean Paul Calderone today which pointed out how useless this is. =)
The problem is basically how asyncore.loop() is implemented.
Being impossible to let loop() function know when a file descriptor gets added or removed from the socket map we are forced to iterate over all existing fds on every loop which nullifies the benefits offered by epoll() and kqueue() syscalls.

As of right now I can't think of a solution to this problem which doesn't imply a change to the current asyncore API.
History
Date User Action Args
2010-04-19 19:55:13giampaolo.rodolasetrecipients: + giampaolo.rodola, josiahcarlson, Ikinoki, lasizoillo
2010-04-19 19:55:13giampaolo.rodolasetmessageid: <1271706913.68.0.421411127812.issue6692@psf.upfronthosting.co.za>
2010-04-19 19:55:12giampaolo.rodolalinkissue6692 messages
2010-04-19 19:55:11giampaolo.rodolacreate