Re: The UR* scheme registry, Citing URL/URI specs

Al Gilman (asgilman@access.digex.net)
Fri, 24 Oct 1997 14:04:38 -0400 (EDT)


From: Al Gilman <asgilman@access.digex.net>
Message-Id: <199710241804.OAA10951@access4.digex.net>
Subject: Re: The UR* scheme registry, Citing URL/URI specs
To: connolly@w3.org (Dan Connolly)
Date: Fri, 24 Oct 1997 14:04:38 -0400 (EDT)
Cc: timbl@w3.org, fielding@ics.uci.edu, masinter@parc.xerox.com,
In-Reply-To: <344FD953.2895@w3.org> from Dan Connolly at "Oct 23, 97 06:10:11 pm"

There are two pieces of this question.  One is a question of
nomenclature: how many names, and what name(s) should we use to
refer to things of the UR* class?

The second is a matter of software architecture: what IETF
documents should the W3C documents for HTML etc. cite, and how?

This message talks to the architecture question.

The HTML specification has trouble deciding which RFC(s) to cite
to define acceptable URLs.  The right answer is that the HTML
spec should not be trying to define acceptable URLs.  This is a
deal between the browser and the marketplace which is somewhat
disciplined through the registry efforts of the IETF, but does
not pass under the thumb of the HTML specification.

The _HTML_ requirements on URLs are that they are expressed in
the appropriate SGML type so that they can be embedded in text
where an attribute value is needed.  What strings (schemes)
browsers should be capable of handling is not the place of HTML
to define.  That is the architecture bug at the moment.  HTML
talks as though it defines URLs by reference to RFCs.  HTML per
se shouldn't be pretending to regulate URLs at all.  The IETF
process should be cited as a source for more information.  This
is an informative, not normative clause, i.e. the more
information available from the IETF is outside the scope of the
HTML language rules.

-- Al Gilman