Re: Attributes in Prelim DAV Spec

At 03:45 PM 11/11/96 -0400, Daniel W. Connolly wrote:
>
>
>But I think there is somthing to it: if we adopt Charlie's
>suggestion, then what about existing URLs with ; in them.
>That is, suppose I've got a document I've been publishing
>as:
>	http://foo.org/docs;id=27
>
>for years, and now I want to add attribute functionality
>to my service. What's the address of the 'author' address
>of the doc above? Is it:
>
>	http://foo.org/docs;id=27;attribute=author
>
>I suppose that could work, but it make the server's job pretty hard:
>the server might have to try lots of different ways of splitting the
>URL between the attribute part and the base resouce part.
>
>Worse yet: what if somebody is already using the name:
>
>	http://foo.org/docs;attribute=XYZ
>
>for something else? Is that a problem?

One ntion that might be applied here is that the server and client would
negotiate the use of the WEBDAV protocol via the HTTP/1.2 Extension
Protocol (PEP). The result of such a negotiation would be an agreement on
certain "standardized" use of various elements of the URI (specifically,
the "params" and the "query" portion). For example, the following URI would
have specific meaning to a server using the WEBDAV protocol extension;

	http://foo.org/docs/adoc;version=current?attribute=DC.author

The server would interpret this as a request for the DC.author attribute of
the current version of the document "adoc"; DC.author is meant to be used
in the same sense as described in "A Proposed Convention for Embedding
Metadata in HTML". One could imagine an HTTP link: header being returned
that refers to the document describing these attributes. One might also
imagine that servers might publish (negotiate) exactly which schema they
were using, along with a URL indicating a machine-readable version of that
schema. Finally, the server might then return a document of type
application/attribute-listing, containing the (machine-readable) values for
the attributes requested for that document.

So, my basic point is that unless the WEBDAV protocol was in effect for
this HTTP interaction, there would be no change to the interpretation of
the URLs specified above. 

--Charlie



----------
Charlie Brooks                           The OpenGroup Research Institute
11 Cambridge Center, 4th Floor           Cambridge, MA 02142
(617) 621 8758                           (617) 225 2943 (FAX)
http://www.osf.org/~cbrooks              c.brooks@opengroup.org

Received on Monday, 11 November 1996 16:10:34 UTC