- From: Kurt Cagle <cagle@olywa.net>
- Date: Tue, 28 Aug 2001 13:34:00 -0700
- To: <xml-dist-app@w3.org>
[Paul Prescod] > That's another way but now you are trading off human-readability which > is a major advantage of both URI's and XML (that's why we don't use > UUIDs and ASN.1!). I think a=b&c=d would be a minimized format. The > serialization would be an expanded one. > Just as there is a canonical XML, canonical Unicode, etc. there should > be a canonical section 5. I've been arguing for this for a long time. Creating a canonical form for requests via a GET message is still the only effective away that low bandwidth devices can employ web services currently, especially if you negate the arbitrary distinction that web services are machine-to-machine content while GET/POST involves human interaction. To me it makes a great deal more sense to provide both an HTTP and SOAP mechanism for making at least some web services calls (HTTP was, after all, intended initially as an RPC mechanism in the first place, if you read the histories), placing the requirement of implementing the dual services on the server vendors. Over time, as new devices are introduced that have SOAP clients built in then you can effectively deprecate the older HTTP architecture once a critical mass of such devices are out there, but otherwise you make the adoption of web services through SOAP limited to a very narrow market, primarily driven by e-commerce. A canonical section 5 definition makes a great deal of sense. It means losing some anonymity of properties -- you have to explicitly define certain properties (the first "method" property is defined as the canonical method, for instance) -- but is this really any different than creating a schema namespace? If you have an existing map, then it becomes a trivial process of mapping this canonical section 5 to a SOAP message for subsequent transmission to intermediaries. -- Kurt Cagle
Received on Tuesday, 28 August 2001 16:33:35 UTC