- From: Joseph Kesselman <keshlam@us.ibm.com>
- Date: Mon, 26 Nov 2001 14:00:47 -0500
- To: Christian Geuer-Pollmann <geuer-pollmann@nue.et-inf.uni-siegen.de>
- Cc: Eric van der Vlist <vdv@dyomedea.com>, XML Signature WG <w3c-ietf-xmldsig@w3.org>, "Joseph Kesselman" <Joseph_Kesselman@lotus.com>
Clarification: The quote re "I'm not sure whether this should be marked as LATER, or as WONTFIX since we expect that XSLT2 will actually make what we're doing legitimate." refers specifically to the fact that Xalan does not replicate the Namespace Node onto every child node -- it does support the namespaces axis, but does so by returning the Namespace Node at the point where the namespace was declared. This difference is essentially invisible unless you absolutely insist on counting the Namespace Nodes present in a subtree or asking for their parents -- both of which are _EXTREMELY_ rare operations in real-world stylesheets and XPaths. Xalan does put a heavy emphasis on standards compliance. But my best information at this time suggests that the XSLT committee has recognized that this concept of replicating the Namespace Nodes onto all the descendents was a Bad Idea in the first place, and that they are planning to remove the ability to ask the questions which expose this difference. If they do so, Xalan will probably be fully compliant as it stands. I'm not sure why this affects anyone writing a canonicalizer. Ideally, you shouldn't care where the namespace was actually declared. Since that doesn't affect the semantics of the document, I would expect canonicalization to suppress that information...? ______________________________________ Joe Kesselman / IBM Research
Received on Monday, 26 November 2001 14:01:23 UTC