W3C home > Mailing lists > Public > www-tag@w3.org > March 2002

RE: The range of the HTTP dereference function

From: Joshua Allen <joshuaa@microsoft.com>
Date: Wed, 27 Mar 2002 14:55:16 -0800
Message-ID: <4F4182C71C1FDD4BA0937A7EB7B8B4C104B615DE@red-msg-08.redmond.corp.microsoft.com>
To: "Mark Baker" <distobj@acm.org>, "Dan Connolly" <connolly@w3.org>
Cc: "Tim Berners-Lee" <timbl@w3.org>, "Roy T. Fielding" <fielding@apache.org>, "www-tag" <www-tag@w3.org>
> I agree strongly with Dan and Roy here.
> What feature would it take in HTTP to allow it to properly support
> identifying cars?  You'd need some sort of assertion declaring a
> relationship between a document and a real thing, right?
> In other words, we'd need a way to declare that
> (which identifies me), and "http://www.markbaker.ca/index.html" (which
> identifies an HTML representation of me, aka my "home page"), were
> related in this manner.

I think the problem is very simple.  If you declare that
http://www.markbaker.ca/ represents the physical you, then you lose the
ability to make assertions about the resource that is returned when you
do a GET on that URI.  If you want to let some URIs do one thing and
others another, you need a way for people (and machines) to figure out
what the "default" sense of that URI is.  The other possibility is to
*always* specify (when recording metadata about a URI) what sense you
are using the URI.  This last option seems like the only safe one to me,
since people obviously are insisting on using HTTP URIs to assign
metadata about things that are *not* GET-able resources.
Received on Wednesday, 27 March 2002 17:56:05 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 22:55:50 UTC