Five minutes on IRC solved it: their libraries aren't threadsafe, and this was one of the most common symptoms of threaded access to their transport.write function. "Undefined behavior" and all that.
So, with a quick reactor.callFromThread(self.transport.write,head+str(data)), this problem is laid to rest! This was, IMO, the #1 bug in pyaim-t. I think this (and, oh yeah, the tremendous patches from Scott Dial and Michael Dyrna) calls for a release! ~Chris Chris Carlin wrote: > I'm trying to track down this longstanding delayed messages bug > associated with pyaim-t. As may have been said before, I can see the > message being handed over to twisted (transport.write), and I can verify > that the traffic is delayed with tcpdump. > > tcpNoDelay is set to true to disable Naugle's algorithm, but that's > about all I can think of to do. It really does look like a buffering > issue, though. > > Do the other transports experience this delay as well? Anyone have any > thoughts on making Twisted behave? > > ~Chris > _______________________________________________ > py-transports mailing list > py-transports@blathersource.org > http://lists.modevia.com/cgi-bin/mailman/listinfo/py-transports _______________________________________________ py-transports mailing list py-transports@blathersource.org http://lists.modevia.com/cgi-bin/mailman/listinfo/py-transports
