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

Re: The range of the HTTP dereference function

From: Roy T. Fielding <fielding@apache.org>
Date: Thu, 18 Apr 2002 15:28:22 -0700
Cc: "Norman Walsh" <Norman.Walsh@Sun.COM>, <www-tag@w3.org>
To: "Sean B. Palmer" <sean@mysterylights.com>
Message-Id: <97473384-531B-11D6-B8E2-000393753936@apache.org>
> Whether it's a document or not, you can refer to it by URI. The question 
> is
> whether the solid object sitting in your hotel room is directly
> identifyable using an HTTP URI.
>
Nothing is directly identifiable using an "http" URI.  HTTP only
provides an indirect interface to resources, even for simple things
like files.  The notion that we can arbitrarily restrict the nature
of what is behind the interface to some preconceived idea of an
implementation is simply, and provably, incorrect.

HTTP only limits the interaction you can have via an identifier,
not the nature of what is being identified.  I can assign an "http"
URI for my girlfriend, but that doesn't mean I want to limit my
interactions with her to HTTP.  Assignment of an identifier does
not in any way imply accessibility via some protocol -- it only
defines the naming authority and an identifier that is significant
to that naming authority.

And no, I don't interact with my girlfriend via HTTP.  She only
accepts requests via SMTP, tel, and a few other less standardized
access mechanisms.

....Roy
Received on Thursday, 18 April 2002 19:03:48 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 26 April 2012 12:47:06 GMT