- From: Paul Libbrecht <paul@activemath.org>
- Date: Tue, 30 Nov 2004 17:05:58 +0100
- To: www-tag@w3.org
Dear TAGgers, I would like to request comments on the following practice which appears to me as good but which seems to be barely done by the current web-services practices: • Following "The Architecture of the Web", a resource is at least something a representation of which can be "GET"(ted). • one should also be able to "operate" on a resource: - Methods on resources could, thus be acted upon using a POST (or something else?). In turn, they can create new resources, returned as URIs, of course. - And some resource representations (that can be "GET"(ted)) should document how this can be done. Doing so, I seem to be (mis)using resources as objects... I am unclear how "evil" this is... This is somewhat similar to WSRF (http://www.globus.org/wsrf/), but WSRF does not declare anything about "GET"ting a resource's representation (and sort of carefully avoids it, I think, because of the GRID's massive data intents). Similarly to WSRF this looks to be very appropriate to describe stateful "web-services" or "resources" (including the ability, probably, to "HEAD" the resource in order to check a state change). Reading about REST services, I would have expected such a practice to have emerged but that seems to have evaporated... even the Semantic Web Services group, whose RESTful-services interest is declared, seemed to be only partially interested by WSRF. paul
Received on Tuesday, 30 November 2004 16:06:41 UTC