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

At 7:08 PM -0500 10/24/97, Tim Bray wrote:
>I don't think Dan was asking me, but

And I don't think you were asking me, but...

>
>1. if the XML spec says URL and somebody sends me a doc with an
>   external reference to urn:ietf:rfc:1661 I probably won't be
>   able to resolve it, but at least I have some self-defense because
>   I can make a strong case that it's not a URL, so I can tell the
>   sender he's not XML-conformant.

>2. If the XML spec says URI and the same thing happens, then I have no
>   defense, because the sender can say "That's a URN, and a URN is a
>   URI, and the spec says I can give you URIs."  I.e. conformance
>   without interoperability.
>
>I think this qualifies as breakage as a direct result of using URI
>rather than URL, but then I'm simple-minded.

Yes, you are being simple minded, because the _same_ problem afflicts your
solution. I can send you a "file:" URL; you can't prevent me without
limiting the legal URL schemes for XML (and "file:" is too useful to
trash). My simple-minded take is that broken references are broken
references, and we _can't_ prevent them, in any kind of addressing scheme.
The fact that your laptop was able to check the wrong filesystem for the
file on _my_ server is rather irrelevant. Broken is broken.

   Location independence is really useful (demonstrated fact). The
counter-argument that location independent names can break is very weak
given that location dependent identifiers also break regularly.

  --  David


_________________________________________
David Durand              dgd@cs.bu.edu  \  david@dynamicDiagrams.com
Boston University Computer Science        \  Sr. Analyst
http://www.cs.bu.edu/students/grads/dgd/   \  Dynamic Diagrams
--------------------------------------------\  http://www.dynamicDiagrams.com/
MAPA: mapping for the WWW                    \__________________________

Received on Friday, 24 October 1997 22:40:35 UTC