I think we got the question right. But I think that rather
than worrying among ourselves, we should contact the XSL
WG and ask them.

Regards,  Martin.

At 00/08/18 14:44 -0400, Ed Simon wrote:
>I think the question to ask is
>"Given a result tree, is it possible to correctly write out
>two different XML streams such that the canonicalized form
>of the streams are different?"
>If the W3C (including ourselves) have done things right, the answer
>should be
>"No.  A result tree has only one canonical form associated with it."
>(Admittedly, I'd like to think more about
>this question but my standards time is kind of pre-occupied on
>the encryption front right now.)
>Just like the IETF interop testing uncovered some subtle parser
>nuances that need refinement, I expect there will be similar
>discoveries with XSLT tools.
>Let's face it, because digital signatures must be absolutely 100%
>unforgiving to any discrepancy (as they must be), bringing digital
>signatures to the world of XML will be a very effective way of
>flushing out bugs in both XML-related specs and the tools built on
>those specs.  Ultimately, that's good for everyone.
>-----Original Message-----
>From: John Boyer []
>Sent: Friday, August 18, 2000 1:47 PM
>To: Martin J. Duerst; XML DSig
>Subject: RE: XSLT
>Hi Martin and Ed,
>Actually, Martin, I do think DSig has some responsibility to document I18N
>limitations of this transform, so we could really use your help on that.
>However, I was more concerned with generic stuff like the fact that when I
>set my XSLT processor to output text/html, it tends to rewrite some stuff on
>me like line breaks and non-minimal minimalized attributes.  I don't
>'serialize the result tree' tree directly as Ed recommends.
>So, I'm fishing for some kind of statement that when an XSLT processor is
>set to output text/xml, that these funny rewrite rules aren't applied to my
>As long as that's true, then a c14n transform after the XSLT should clean up
>the mess; I just want some additional assurance as we approach CR that we
>aren't messing this up. (And, again, thanks for your patience, Ed).
>John Boyer
>Development Team Leader,
>Distributed Processing and XML
>PureEdge Solutions Inc.
>Creating Binding E-Commerce
>v: 250-479-8334, ext. 143  f: 250-479-3772
>1-888-517-2675 <>
>-----Original Message-----
>[]On Behalf Of Martin J. Duerst
>Sent: Thursday, August 17, 2000 7:54 PM
>To: John Boyer; XML DSig
>Subject: Re: XSLT
>At 00/08/17 12:30 -0700, John Boyer wrote:
> >I've often wondered, how open-ended is XSLT regarding the output.
> >
> >Are there permissible implementation differences that are not covered off
> >by throwing a c14n transform after (or before and after)?
>There are definitely things that fall in this category, in particular
>in the area of localization. For example you can say that you want
>some items sorted, and can indicate that the sort order should
>be according to e.g. Swedish practices, but you are not guaranteed
>that your processor knows Swedish sorting, nor are you sure that
>all processors do exactly the same for Swedish even if they know it.
>This is I guess quite a bit away from your main problem.
>Regards,   Martin.
> >Or rather, is there an XSLT conformance mode that guarantees any
> >implementation adhering to that mode produces the exact same output
> >(except possibly for differences that can be corrected by c14n, and
> >possibly only if the input is canonicalized)?
> >
> >Or is it the case, for example, that random extra whitespace may be added
> >outside of start tags by some processors and not by others?
> >

Received on Saturday, 19 August 2000 05:19:27 UTC