W3C home > Mailing lists > Public > www-tag@w3.org > June 2005

RE: [schemeProtocols-49] Editor's draft of finding on schemeProtocols-49

From: Larry Masinter <LMM@acm.org>
Date: Fri, 17 Jun 2005 12:42:55 -0700
To: "'Mark Baker'" <distobj@acm.org>
Cc: www-tag@w3.org, noah_mendelsohn@us.ibm.com
Message-id: <0II800L06UROZH@mailsj-v1.corp.adobe.com>

Let's suppose I have a speed limit sign, "Speed Limit 55 mph".

Most people drive 64 mph, however, because they know that
if you're driving less than 10 mph over the speed limit,
you'll unlikely to get a ticket.

Now, the question is: what does the sign MEAN? Does it
mean "Speed Limit 55 mph", or does it really mean
"Speed Limit Really 64 mph".

Well, you could ask the person/government that put up
the sign, what did they mean? Did they really want people
to drive 55? Or did they really MEAN that it was OK to
drive 64?

I think that if you're going to write a standard about
what the sign means, the "standard" meaning of the
sign should match literally what the sign says, independent
of the true intention of the person who put up the sign,
or the behavior of those who read and follow the sign.

A URI is a sign. The fact that it says "ftp://ftp.w3.org/blah"
signifies that it's *meaning* is "Use the FTP protocol".

Even if the person who writes the URI (puts up the sign)
really knows that many people use a handy FTP-HTTP proxy,
and that they know that he knows it, and even would rather
prefer that they use the proxy rather than FTP directly.


> I disagree.  I believe that the relationship between URIs and the
> resources they identify is independent of URI scheme and protocol.  Your
> examination into gateways seems to demonstrate this, as, for example, a
> resource identified with an ftp URI is able to be interacted with
> via an HTTP proxy in a manner indistinguishable from a black box POV,
> from an http URI (see below).  Even a protocol as mismatched with HTTP
> as TELNET, can be interacted with via the GET semantic, and is
> indistinguishable from double-clicking on a hostname in your telnet/SSH
> client (at least for the initial interaction that establishes the
> session).
Received on Friday, 17 June 2005 19:44:02 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 26 April 2012 12:47:36 GMT