W3C home > Mailing lists > Public > uri@w3.org > April 2004

Re: test cases and implementation reports

From: Paul Grosso <pgrosso@arbortext.com>
Date: Tue, 20 Apr 2004 14:49:51 -0500
Message-Id: <4.3.2.7.2.20040420143844.01987b20@172.27.10.30>
To: Martin Duerst <duerst@w3.org>, "Roy T. Fielding" <fielding@gbiv.com>, uri@w3.org

At 16:59 2004 04 20 +0900, Martin Duerst wrote:

>At 16:23 04/04/20 +0900, Martin Duerst wrote:
>
>>I have created a testing page at http://www.w3.org/2004/04/uri-rel-test.html
>>to make (manual) testing of the implementations in various browsers easy.
>
>Here is another test, for Opera (V7.23, build   3227, Win2000):
>
>All tests are passed, except the following:
>
>#       rel             correct                 Amaya
>
>6       //g             http://g                http://g/ (additional slash)
>9       #s              http://a/b/c/d;p?q#s    {testpage URI}#s

This goes back to my earlier email on "intra-document" URI references.  
I thought I finally understood Roy's explanation, but I guess I never did.

I kept saying that a "fragment identifier only" relative URI
reference should reference the current resource regardless
of the base URI, and Roy kept giving answers that I didn't 
understand until John Cowan finally convinced me Roy's words
were a different way of saying the equivalent thing.

But my (and Amaya's and Opera's) understanding of RFC 2396 is
that #s should resolve to {testpage URI}#s, so if Roy's rewrite
of RFC 2396bis now has #s resolving to http://a/b/c/d;p?q#s
when the #s relative URI reference lives in another resource,
then I still think this is a change in semantics to RFC 2396,
and one that breaks things (such as the several examples I
gave in the earlier thread).

paul
Received on Tuesday, 20 April 2004 15:53:15 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:25:07 UTC