- From: Mark Baker <distobj@acm.org>
- Date: Thu, 25 Oct 2001 12:39:13 -0400 (EDT)
- To: Noah_Mendelsohn@lotus.com
- Cc: marc.hadley@sun.com (Marc Hadley), kumeda@atc.yamatake.co.jp (Kumeda), ms@mitre.org (Marwan Sabbouh), skw@hplb.hpl.hp.com (Williams Stuart), xml-dist-app@w3.org
> > Mark Hadley writes: > > >> The application programmer shouldn't care > >> what the underlying protocol is, only that > >> it supports the semantics > >> required by the application. e.g. if > >> the application only requires a > >> request-response message exchange pattern > >> then it would work over any binding that > >> supports request-response. > > +1. Exactly. -1/2. This is only appropriate in the tunnel view of the underlying protocol. In the "Chameleon View" (good name, Stuart), this is not the case. Or to put it another way, if your underlying protocol is an application protocol, then I agree with Marc when he writes ".. only that it supports the semantics required by the application". Because that's what an application does; it defines application semantics. MB -- Mark Baker, CSO, Planetfred. Ottawa, Ontario, CANADA. mbaker@planetfred.com
Received on Thursday, 25 October 2001 13:20:23 UTC