Lists Home |
Date Index |
At 19:38 21/01/2002 +0000, Miles Sabin wrote:
>Jonathan Borden wrote,
> > It seems to me that the benefit of HTTP is that it allows the
> > 'owner' of the URI, which means the registrant of the DNS entry of
> > the hostname, or owner of the IP address, either of which is the
> > "host" part of the URI, to make some statement regarding what the
> > URI is intended _by the owner_ to mean.
>Who cares what the owner means. There's nothing to stop anyone else
>using that very same URI to mean something else. That's why URIs can't
>on their own convey meaning. To get that you need semantic agreement
>between the producers and the consumers of the URIs.
Um, this is unduly pessimistic.
You are right, of course: there's absolutely nothing to stop me using the
<html:p> element to mean "transfer lots of money to my bank account".
Indeed, if you and I had a private agreement to use it that way, we could
build a perfectly workable system that way.
But XML is often used in a publishing context, rather than a point-to-point
messaging context, and explicit private semantic agreements are neither
possible nor necessary. A tacet agreement to use the meanings defined by
the owner of a namespace is perfectly workable.