Message-Id: <9306040700.AA20294@mocha.bunyip.com>
Date: Fri, 4 Jun 93 3:00:02 EDT
From: John Curran <jcurran@nic.near.net>
To: uri@bunyip.com
Subject: TYPE information in URLs
Folks,
If we presume that the URN refers to the intellectual content of
the resource (which may representated in several formats such as text,
postscript, and rtf), then it's clear than associating ONE format identifier
with a given URN will not work.
Yes, it is possible to associate the format id with the URL, if we presume
that a URL refers to particular instance, and that a particular instansiation
can only have one actual format. With these assumptions, it becomes possible
to encode optional format in the URL, if we so desire.
The format identifier is one of several resource instance attributes that
could be encoded in optional URL fields; others include: size in bytes, cost
to access, access control information, and more. The problem is that this
information can change independent of the URL, and leave us with incorrect
values embeded in URL's.
Many of the protocols (wais, gopher+, www) have application-specific
methods for passing with the additionnal information, and we should not try
to standard such identifiers until we have solid operational experience.
At such time, we can reaevaluate our needs and determine if an augmented URL
or a seperate [URL, URx] pair would serve us better.
/John