http://www.jacek.cz/blog/comments.xml 2005-02-01T23:44:29Z Movable Type 3.31 http://www.jacek.cz/blog/archives/2005/02/processmessage/#comment-177 2005-02-01T23:38:39Z 2005-02-01T23:38:39Z Hi Jacek. So far I haven't seen anything that backs up your statement that "ProcessMessage represents a way for a sender of a message to be sure the recipient wil do what the sender says". All it says is that a receiver gets a message; what it does with that message is implementation specific. OK, you'd kinda hope it would do something based on the content, but it could just junk it, perform some internal transformation on it, or anything. ProcessMessage doesn't say what the implementation does at all.

]]> Dave Corne http://www.jacek.cz/blog/archives/2005/02/processmessage/#comment-178 2005-02-01T23:44:29Z 2005-02-01T23:44:29Z ProcessMessage isn't a useful architectural approach. It's too low level in one manner and too darn obvious in another - everything is based on messages. Hey, why don't we retrofit this to COM, DCE, CORBA and even J2EE?! They'll all map afterall. It just doesn't say anything about loose-coupling that it couldn't easily say for close-coupling too. But pub-sub is different; it's a naturally understood paradigm that's worked well for decades.

]]> anon