Re: FW: LC Comments: Web Method Feature

On Thu, Jul 04, 2002 at 11:45:15AM +0100, Marc Hadley wrote:
> I think this is an instance of the top-down vs bottom-up dichotomy. In 
> the top-down view the app/developer would choose a suitable MEP and the 
> binding would deal with the HTTP-specific stuff like choice of HTTP 
> method. I.e. the framework would offer a clean separation between the 
> application and the underlying protocol.

But the protocol is the application.  The methods in an application
protocol are the same kind (same layer) as the methods seen in an API.

The cleanest separation you could offer - and I've had this discussion
with Noah before - is to define the semantics of an application protocol
independantly of the specification of that protocol.  So we might
extract all of the application semantics out of RFC 2616, place them in
some new spec, and then be able to build new HTTP-like protocols with
that spec (say, that are optimized for certain environments - not that
I'm suggesting this is a good idea).

But that would still require the developer to choose the method.
Perhaps my last response to Stuart will help explain that better.

> Is that any clearer ?

Yes, thanks.

MB
-- 
Mark Baker, CTO, Idokorro Mobile (formerly Planetfred)
Ottawa, Ontario, CANADA.               distobj@acm.org
http://www.markbaker.ca        http://www.idokorro.com

Received on Thursday, 4 July 2002 12:12:04 UTC