WebArch and RFC2396bis: URIs and Fragids

This is a Last Call comment to the Web Architecture Document.
Because it potentiall affects RFC2396bis, I'm also copying
the uri mailing list.

The comment is mainly about Section 3.3, entitled "Internet
Media Type" (as most of the section is about fragids, it may
be better to change the title, or move most of that material
to 2.6).

This comment is about streamlining terminology. For URIs, a
clear distinction is made between Resource and Representation.
But when it comes to fragid, this distinction is suddenly gone,
without any explanations.

After reading the Web Arch doc, which helped me think through
these issues, I got to the following table:


Identifier   Abstract     who defines   Concrete
                             mapping

URI          Resource     scheme:     Representation(s) of Resource
URI#fragment Subresource  mime/type   Representation(s) of Subresource


In words:

An URI identifies a Resource. To get a representation of the resource,
you start with the scheme:, which tells you how you (potentially) get
a representation.

An URI#fragment identifies a Subresource. To get a representation of the
subresource from a representation of the resource, you start with the
mime/type of the representation of the resource, which tells you how you
(potentially) get a representation of the subresource.

This would allow to bring more consistency (sic!) to the Good Practice
points. Currently we have:

Fragment identifier consistency: to resource owners, keeping URIs with
     fragids consistent across content negotiation)
     (there is no such point for URIs, although this of course applies
      very much so to URIs, too)
URI ambiguity: This seems to correspond to the above, or subsume it,
     but the terminology used is quite different
Consistent URI usage: this is about not changing a URI

The language for the definition of fragid semantics could then also
be modified. Currently, there are things like "The Internet Media
Type of the retrieved representation specifies the authoritative
interpretation of the fragment identifier.". Why not just say
something like "The Internet Media Type specifies how to obtain
the Representation of the Subresource from the Representation of
the Resource."



Regards,   Martin.

Received on Monday, 1 March 2004 03:40:05 UTC