RE: Don't put descriptive info in a URI, only identifying info

From: Larry Masinter (masinter@parc.xerox.com)
Date: Tue, Dec 22 1998


From: "Larry Masinter" <masinter@parc.xerox.com>
To: "Michael A. Dolan" <miked@tbt.com>, "Roy T. Fielding" <fielding@kiwi.ics.uci.edu>
Cc: "Dan Connolly" <connolly@w3.org>, <ietf-url@imc.org>, <www-tv@w3.org>
Date: Tue, 22 Dec 1998 15:45:37 PST
Message-ID: <003901be2e05$2d0ea040$aa66010d@copper.parc.xerox.com>
Subject: RE: Don't put descriptive info in a URI, only identifying info 

Think of the 'news' URL scheme, and our (unsuccessful) attempt to
separate out 'news' and 'nntp'. We wound up with one scheme which
was used for identifying a variety of types of resources:
A stream of information (like a channel):

   'news:comp.infosystems.www'

A particular message (like an event):

   'news:74fqt5$ge7@arcade.netmar.com'

In addition, the 'news' scheme was used for NTTP:

    news://news.parc.xerox.com/comp.infosystems.www.announce

(news.parc.xerox.com is a Xerox-internal nntp server.)

There were other elements of the news stream for which there
are no URLs, even though URLs would be 'useful', e.g., 
'a conversational thread'.

So it's not as if we have no experience with other kinds of
'streaming broadcast' material in the URL space. It's not clear
that the 'news' URL scheme is a wonderful example that must
be followed, but it's at least prior art.

In the netnews space, at least, articles have unique IDs that
allow them to be identified independently of the stream in 
which they are broadcast. That isn't the case for most of TV
today (is it?) so you may need more designators, but don't try
to pack an arbitrarily complex expression into the locator.
The scenarios will help figure out what's truely useful.

Larry
-- 
http://www.parc.xerox.com/masinter