Re: uri-comp draft necessary?

Am Mittwoch, 18.12.02, um 02:30 Uhr (Europe/Berlin) schrieb Tim Bray:

>
> At several points during yesterday's TAG call, Dan claimed repeatedly 
> that URIs a and b are equivalent if (strcmp(a,b)==0) equal, otherwise 
> not, end of story.  If we believe this, we can discard the uri-comp 
> draft entirely right now and I won't put any more work into editing 
> it.  Do we believe this? -Tim
>
As the draft itself points out, there are applications such as browsers 
and crawlers which
need better equivalence than strcmp. RFC 2396 says that scheme names 
are case insensitiv, so
strcmp produces false negatives, without even going into scheme 
specific comparisions.

As a small example of strcmp shortcomings: XML namespaces seem to work 
fine with their
definition of equivalence. However the current efforts to make 
"namespace documents"
accessible on the web will have to cope with the false negatives of 
strcmp. One will
have a hard time providing different documents for 
"http://example.org/myns" and
"http://EXAMPLE.ORG/myns".

//Stefan

Received on Wednesday, 18 December 2002 04:34:17 UTC