Re: Documents, Cars, Hills, and Valleys

[snip]

I'm inclined to agree. SOAP Web service endpoints are another good example
of http:-named resources for which the 'document' metaphor seems to be
over-stretched.

There's a large class of http:-named resource for which the resource can
be exhaustively described by a content-typed bag of bytes. But there are
other resources (typically those that wrap databases, services etc) for
which we can never get a complete rendering of 'the thing itself', only
exchange messages with it. The document metaphor(*) seems too passive a
notion to effectively characterise such resources, yet in practice we
deploy http: names for them without much trouble.

Dan


(*) and it _is_ a metaphor; we shouldn't get too hung up on arguing
whether a Web service really 'is' a kind of document, since we're mostly
testing the limits of our intuitions about the word 'document' here,
rather than probing anything interesting about the world.

Received on Wednesday, 10 April 2002 16:14:40 UTC