- From: Mark Baker <distobj@acm.org>
- Date: Tue, 20 Nov 2001 11:55:46 -0500 (EST)
- To: moore@cs.utk.edu (Keith Moore)
- Cc: brian@hursley.ibm.com (Brian E Carpenter), discuss@apps.ietf.org (Discuss Apps), john_ibbotson@uk.ibm.com, rpk@us.ibm.com
I agree with Keith. If you're tunneling over HTTP, i.e. using it as a transport protocol, then perhaps this is a useful thing to do, I don't know. But as HTTP is an application protocol, "reliability" must be interpreted in the context of the application it defines. This draft doesn't do that. > overall this is a good requirements summary. > > But I find no justification whatsoever for insisting that the reliable > message delivery system be layered on top of HTTP. One could easily > conclude that expectation that HTTP should be used as a substrate for > applications requiring reliable message delivery is a large part of > the problem. > > Keith MB -- Mark Baker, CSO, Planetfred. Ottawa, Ontario, CANADA. mbaker@planetfred.com
Received on Tuesday, 20 November 2001 11:50:57 UTC