- From: Florent Georges <fgeorges@fgeorges.org>
- Date: Sat, 29 Nov 2008 20:32:20 +0100
- To: "Dave Pawson" <dave.pawson@gmail.com>
- Cc: public-xml-processing-model-comments@w3.org
2008/11/29 Dave Pawson wrote: > 2008/11/29 Florent Georges <fgeorges@fgeorges.org>: >> But instead of telling about extensions (besides, you don't tell >> about XProc extensions, but XSLT extensions...) you can maybe relax >> the MUST and tell it is implementation -defined or -dependent. > Which is what XSLT does? Except XSLT has no reason to mention > extensions? I.e. they are always implementation dependent? I would say that the situation is different. XSLT 1.0 does mention extensions, and even define two kinds of extensions (functions and instructions) and sets some rules for them. But XSLT 1.0 does not mention final result trees, because the result in XSLT 1.0 is the result tree (only one.) So to summarize: XSLT 2.0 has (possibly) several result trees while XSLT 1.0 has only one. So the first thing in the draft was to say that "in XSLT 1.0 additional result trees MUST be the empty sequence." But well-established XSLT 1.0 extensions (for instance from EXSLT, Xalan and Saxon 6) provide a similar semantic for XSLT 1.0 as well. So clearly, the MUST has to be removed. But I think that mentioning XSLT extensions in XProc sounds over-specified. -- Florent Georges http://www.fgeorges.org/
Received on Saturday, 29 November 2008 19:32:56 UTC