W3C home > Mailing lists > Public > xml-uri@w3.org > May 2000

Re: Irony heaped on irony

From: <keshlam@us.ibm.com>
Date: Thu, 25 May 2000 09:59:10 -0400
To: "David Brownell" <david-b@pacbell.net>
cc: xml-dev@xml.org, xml-uri@w3.org
Message-ID: <852568EA.004CD16D.00@D51MTA03.pok.ibm.com>
>Well, there IS a specification (the URI document) that talks about
>relative URIs ... it's not a W3C document, it's an IETF one.

I believe that was phased out in favor of moving relative syntax into URI
References.

>Issues about "%NN" escaping crop up too, but again it's clear (IMHO) that
>such escaping is for encodings of the URIs, not to the URIs themselves.
>So again, comparing URIs must remove such escapes.

This is one distinction between Absolutized and Canonicalized form;
character escapes are _not_ expanded in the former, according to the
algorithm given in the URI spec. This point of confusion is one reason some
of us are dubious about whether Absolutizing is really the right answer
even if one did want to "compare the names as URIs."

>It also appears that some folk have decided to use the namespace
comparison
>bug (above) as a tool to restart that dereferencing debate.  That really
>doesn't help lead to resolutions.

I agree with you, since dereferencing would be behavior of higher-level
code rather than behavior of the namespace spec itself -- and since
dereferencability (and URIRef binding without dereferencing) of the
namespace can be accomplished through other mechanisms even if the
namespace name was in a completely not-web-like syntax.



I know, I haven't responded to TimBL's comment about chemical plants
blowing up -- because I haven't had a chance this week to work my way back
that far in the discussion. However, I'm very confused by the concept -- I
can certainly see arguments for being able to bind a namespace to a URIRef,
but I still can't see any reason that doing so explicitly, rather than
implicitly via the Namespace Name, would make _ANY_ semantic difference in
the processing of a document.

It's certainly possible Tim has thought of something I haven't; I'll
comment again after I've read his use case. But my point of confusion
really is that the only differences I see between
     <foo:bar xmlns:bar="URIRefToBeAbsolutized"/>
and
     <foo:bar xmlns:bar="JustANameTakenLiterally"
          xmlns-binding:bar="URIRefToBeAbsolutized"/>
are that the latter is a bit wordier, takes an extra few cycles to process
when you really do want the binding -- but is a heck of a lot more clearly
defined and straightfoward.

If implicit behavior causes confusion, making it explicit is always a good
bet.
Received on Thursday, 25 May 2000 09:59:44 GMT

This archive was generated by hypermail 2.2.0 + w3c-0.30 : Tuesday, 12 April 2005 12:17:22 GMT