Re: get metadata (was RE: WS-Addr issues

Hi David,

On Thu, Nov 04, 2004 at 09:07:23AM -0800, David Orchard wrote:
> 
> Mark,
> 
> There's a big difference between GETting the state of a resource (which
> may or may not be a metadata resource), GETting the metadata of a
> resource (which may or may not be a metadata resource), and GETting the
> state of a metadata resource.

Well, as you've done there, all are expressable in terms of "GET".
There are differences certainly, but I don't see why those differences
must be manifested in the interface ...

>  There are only 2 solutions to GET
> metadata about a resource that you don't know is a metadata resource:
> either have 2 different operations (GET state and GETMETADATA state) or
> have mechanism to convert an identifier into a metadata URI (URI +=
> "?WSDL" ? ).  

I count at least four.  Add;

- enable a representation of a resource to tell you that it's the
metadata for some other resource

- enable a representation of a resource to tell you that some other
resource is its metadata resource

I was talking about the latter when I suggested a standardized
"metadata" term, but the former may also have value.  I don't believe
either of those two options you outlined are particularly attractive
because I don't believe the "meta" relationship is special.  Consider
that WebDAV once thought that "properties" were special and so minted a
new HTTP method called PROPFIND.  But that's now generally regarded as
a mistake, and that GET should have been used instead.

> To a great deal, http-range-14 is TimBL's attempt to make all http:
> resources into metadata so as to avoid the 2 verb problem for the
> semantic web.

Interesting perspective, but I disagree.  Tim's very much a proponent of
a single "gimme data" method, and I never saw that questioned during the
httpRange-14 debates.

Mark.
-- 
Mark Baker.   Ottawa, Ontario, CANADA.        http://www.markbaker.ca

Received on Thursday, 4 November 2004 17:35:02 UTC