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

Re: RFC 2396bis sec. D.2 editorial suggestion

From: Mike Brown <mike@skew.org>
Date: Thu, 15 Jul 2004 04:02:26 -0600 (MDT)
Message-Id: <200407151002.i6FA2QZk044100@chilled.skew.org>
To: uri@w3.org

Roy T.Fielding wrote:
> I appreciate the reordering of sentences, but most of your description
> is wrong -- it has no such effect on 2396 same-document references
> because those references consisted only of fragment identifiers.

Ah, you're right, of course. The problem is that I have two pieces of 
information that somehow merged into one:

1. When reading about RFC 2396bis (either in the list archives, the drafts, or 
the issues log), I somehow gleaned that the Way Things Are Going To Be is in 
fact the Way Things Were Meant To Be All Along; i.e., any clarifications in 
2396bis are / will be retroactive. (True, yes?)

2. XSLT's document() function resolves its argument(s) to URI references and 
then to a URI against a known base. It is required to return the same node 
(with the same ID) for the same resolved URI. Thus, if implemented properly,
it is essentially doing RFC 2396bis same-document determination.

A year ago, I had gone ahead and implemented the latter in an XSLT processor 
by making its URI resolver to make same-document determinations, which it 
wasn't doing at all before (previously, it relied entirely on the RFC 1808-era 
resolver that comes with Python). Later, as I started making actual RFC 2396bis
updates to the URI library, I probably just assumed that everything that was
in there already was based on 2396. But yeah, now that I think about it, it
was based on XSLT 1.0. Sorry about that.

In any case, your rewritten version looks great; thanks.

-Mike
Received on Thursday, 15 July 2004 06:02:30 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 13 January 2011 12:15:34 GMT