- From: Hammond, Tony (ELSLON) <T.Hammond@elsevier.com>
- Date: Wed, 25 Feb 2004 11:39:57 -0000
- To: 'Larry Masinter' <LMM@acm.org>, "'Williams, Stuart'" <skw@hp.com>, "'Roy T. Fielding'" <fielding@gbiv.com>
- Cc: uri@w3.org, 'Graham Klyne' <GK@NineByNine.org>
Hi Larry:
> I think fragment identifiers are only defined for use with retrieval,
> because the semantics of the fragment are (supposed to be, at least)
> defined by the media type of the retrieved result. With other
> operations,
> there is no clear media type.
I wonder what this means with respect to the INFO URI scheme which defines a
fragment component as a regular part of the syntax,
info-URI = info-scheme ":" info-identifier [ "#" fragment ]
while at the same time asserting that INFO URIs are non-dereferenceable (and
hence there are no representations and no associated media types). In
discussing the use of fragment components in the INFO I-D we use language
which is closely aligned with that used in 2396bis (sect 3.5) to assert
that:
The (unescaped) values for the "fragment" component identify
secondary
information assets with respect to the primary information asset
which is referenced by the "info-identifier".
It would seem that the historical use of fragment components has been to
provide an addressing mechanism into a resource representation following a
retrieval operation, as commonly used in HTTP requests. But in a pure
information context there may well be non-dereferenceable URIs such as INFO
which still have a clear need to articulate secondary resources with respect
to primary resources. So I do query what the role of media type is in these
contexts.
Tony
Received on Wednesday, 25 February 2004 06:40:02 UTC