Hi Mark, > Don't underestimate the complexity of writing > asynchronous IOCP based applications in *any* language - just because > something is written in C# and *only* targets one way of handling sockets > doesn't make it the best or most reliable solution.
No doubt even given the best mechanism, the complexity and programmers skill can bring it to naught, but since my team will be writing a solution from "start", we have to evaluate the best that the OS supports. That is also why Twisted looks so promising, not that it has downright IOCP support but at least from what i read the twisted team is working on it and there are already a reactor with iocp and the framework looks easy. > I'm not quite with you here, but I'll give you the benefit of the doubt > and > assume it wasn't your intention to be rude to someone who has been trying > to > help you. --I am not familiar with how Twisted is implemented in Windows, if it uses the --specific windows apis and asynch model winsock2 directly for optimal --performance. -The COM was -optimal for windows because its asynch api winsock2 and -that beats berkeley socket "select" anyday on Windows, but your hint on the -IO completion port gives twisted an equal footing in terms of raw -efficiency. (the iocp support for sockets is in winsock2...) As you can see Mark, there was never a discussion on blocking sockets vs Select. So i wasn't expecting Tim's reply which i won't be posting here. As i said, lets close the thread. Tim is one of those people that this mailing list needs, i can see he responds quickly to help others while ppl like me would just post questions. Yes i do apologize for using the term "planet megatron" . :-) > > Regards, > > Mark > _______________________________________________ python-win32 mailing list python-win32@python.org http://mail.python.org/mailman/listinfo/python-win32
