Re: AW: Poll: Relative URIs and Strings in xmlns attributes

At 10:35 9/21/2000 +0200, Gregor Karlinger wrote:
>My way of thinking is that an algorithm performing Canonical XML must return
>an error if it dedects relative URIs in the input. Maybe I am wrong, Joseph?

My intent was to think that there would be an error. However, I hadn't 
considered the idea of saying there is a canonical form, it just won't 
necessarily be consistent/interoperable. This sounds like a reasonable 
interpretation of the plenary decision, but then it seems we aren't doing 
anyone a favor. If we're going to "permit" relative URIs when we should be 
encouraging their deprecation, at least we should do it consistently...?

(More later).

BTW: I understand our treatment of this issue was briefly discussed at the 
XML Plenary this week with the following two sorts of positions

>...the issue of the c14n spec defining a canonical form for
>         <aDoc xmlns="../foo"/>
>
>         - it's against the plenary decision
>                 (from Michael Sperberg-McQueen, among others)
>
>         - it seems reasonable, even given the plenary decision
>                 (from Noah Mendelsohn, among others)


__
Joseph Reagle Jr.
W3C Policy Analyst                mailto:reagle@w3.org
IETF/W3C XML-Signature Co-Chair   http://www.w3.org/People/Reagle/