RE: AW: Call for Implementation: Canonical XML Becomes a W3C Candidate Recommendation

Hi Joseph and Gregor,

I was pretty sure we covered off the correctness of Example 3.7 last time.
There is no justification in any of the recommendations for ignoring the ID
type of e3's attribute when using a non-validating parser.

Though the change is simple, I would prefer not to change the example at
this time since it would also require a revalidation effort among the other
implementers who had no problem with this.

As well, your assertion that the Xerces DOM parser cannot select by id
unless using validation seems, on the surface, to contradict IBM's
interoperability report since they are most certainly using Xerces and yet
were able to complete example 3.7.

Moreover, if the id() function is something you have to implement, it is not
actually very hard to implement as long as 1) the parser correctly types the
id attribute when not validating, and 2) it is easy to hook your resulting
id() function into the Xpath implementation that evaluates the expression
given in example 3.7.  Are you saying that one of these two things isn't
working.

And again, if you are, then perhaps we should talk to Kent to find out how
IBM got it to work-- because it is certainly a legal example.

Thanks,
John Boyer

-----Original Message-----
From: w3c-ietf-xmldsig-request@w3.org
[mailto:w3c-ietf-xmldsig-request@w3.org]On Behalf Of Gregor Karlinger
Sent: Thursday, November 09, 2000 11:52 PM
To: Joseph M. Reagle Jr.; Gregor Karlinger
Cc: John Boyer; IETF/W3C XML-DSig WG; Petteri.Stenius@done360.com
Subject: AW: AW: Call for Implementation: Canonical XML Becomes a W3C
Candidate Recommendation


Hi Joseph,

> At 11:22 11/3/2000 +0100, Gregor Karlinger wrote:
> >Please fill in a "Y" in all fields of the matrix,
>
> Ok, have done [1]

Thanks.

> >* Example 4 does not result in the expected canonicalization, since the
> >   XML parser used does not correctly normalize an attribute of type ID.
>
> Ok, I presume the parser will be fixed or tweaked. (Note I
> believe both of
> these problems were encountered by Petteri as well [2].)

Yes, I have heard that there will be a fix in the Xerces parser we are
using in the next update.

> >* Example 7 only results in the expected canonicalization, if the test
> >   file is modified in a way, that the example can be parsed using a
> >   validating parser. Otherwise the id function in the document subset
> >   selector XPath will not return any element.
> >John,
> >a question regarding example 7: Can I really expect from a non validating
> >parser, that it recognizes the types of attributes, especially the type
> >of ID attributes? If not, then an XPath using the id() function cannot be
> >used for selecting a document subset to be canonicalized.
>
> I'm not sure how this thread ended, did John's response [3] close
> the issue,
> are we still thinking, or actually disagreeing about adding "the
> remaining
> parts of the DTD, and additionally provide some textual description about
> this problem of using the id() function. [4]

Good question. Having read [3], I also thought John would like to close the
issue.

I am not very happy with the current situation, since I do not want to build
the functionality for the XML id() mechanism for my own, but want to utilize
existing parsers.

For the Xerces DOM parser I can state, that selecting a
DOM element by specifying the value of its ID attribute is only possible if
the DOM model has been built using validating parsing.

Regards, Gregor
---------------------------------------------------------------
Gregor Karlinger
mailto:gregor.karlinger@iaik.at
http://www.iaik.at
Phone +43 316 873 5541
Institute for Applied Information Processing and Communications
Austria
---------------------------------------------------------------

Received on Friday, 10 November 2000 16:11:32 UTC