[medusa] Re: Hideously long selects?

[Sam Rushing]

skip@m... writes:
> [ long timeout ramblings deleted ]
> 
> I tried shortening the select timeout to 2.0 seconds. Same problem. 

My guess is that your accept()ing socket is getting closed. In this
kind of situation I usually add a 'print r' line just before the call
to select(). This will tell you exactly which objects are waiting for
read. If your server is not in there, then you have your culprit.

> When I made the change I noticed that I picked the wrong version of
> asyncore.py to modify. When I compared them I got the following
> version info:
> 
> The version from ZServer had:
> $Id: asyncore.py,v 1.2 1999/04/09 00:37:33 amos Exp $
> 
> The Python (1.5.2+) version had:
> $Id: asyncore.py,v 1.2 1999/06/08 13:20:05 guido Exp $
> 
> Any chance I'm seeing some version skew problems?

Hmmm.. I get:

$Id: asyncore.py,v 2.44 1999/07/27 00:50:06 rushing Exp $

8^)

We probably have three completely different CVS repositories at work
here.

To get the latest version from medusa:

bash$ export CVSROOT=:pserver:medusa@s...:/usr/local/cvsroot
bash$ cvs login ['medusa' is the pa**w**d]
bash$ cvs checkout medusa/asyncore.py

-Sam