how URIs take on meaning (2.6)

I have never been quite comfortable with...

"   1. The authority over an absolute URI reference determines which
resource it identifies."

http://www.w3.org/TR/2002/WD-webarch-20020830/#uri-generalities

... but until just recently I wasn't sure how
to improve it.

I think it's pretty well established that
names take on meaning by use. With that
background, I suggest something ala:

The URI spec[RFC2717,RFC2396] is an agreement about how
the internet community allocates names and associates them with
protocols by which they take on meaning; for example, the HTTP
URI scheme[RFC2616*] uses DNS in such a way that the
names such as http://somedomain/somepath#someFrag take on
meaning by way of messages from
the domain holder (or somebody they delegate to).
While other communications (documents, messages, ...)
may suggest meanings for such names, it's a local
policy decision whether those suggestions should
be heeded, while the meaning obtained thru
HTTP GET is, by internet-wide agreement, authoritative.

*section 3.2.2 in particular
http://www.w3.org/Protocols/rfc2616/rfc2616-sec3.html#sec3.2.2


Hm... there's probably more to say... the GET response
can be in a media type that allows links/references,
and in that case, linked documents contribute to
the meaning of the term in an authoritative way.

e.g. <http://somedomain/somepath> might be an
XML Schema, and it meaning might depend on included
schemas, and even if those schemas aren't controlled
by the same domain holder, they're part of the
meaning of the URI by internet-wide agreement.

I think TimBL observed, in the recent ftf, that
we don't have a section on how the web is
self-describing or on grounded documents or
whatever. Maybe something to think about in
section 3 on Formats?

Yes, I think it came up in the discussion of formats,
though it seems to have been the sort of off-hand
remark that didn't get recorded.
http://www.w3.org/2002/09/24-tag-summary#formats

-- 
Dan Connolly, W3C http://www.w3.org/People/Connolly/

Received on Thursday, 24 October 2002 10:42:14 UTC