RE: [xsl] An (almost) identity stylesheet

Subject: RE: [xsl] An (almost) identity stylesheet
From: "Michael Kay" <mike@xxxxxxxxxxxx>
Date: Thu, 10 Apr 2008 13:07:56 +0100
>why was that choice made?

It was considered unreasonable to require an XSLT 2.0 processor to know
about and enforce all the XSLT 1.0 rules as well as the XSLT 2.0 rules. This
would have to include semantic differences (e.g. the distinction between
RTFs and nodesets) as well as syntactic restrictions, which could seriously
distort an implementation. Also, it doesn't help migration. Users want to
move forwards to take advantage of new features, while still getting
backwards compatibility mode for their existing code. If they just want a
pure XSLT 1.0 processor, they can use one.

(That's a justification of the choice, not a historical account of WG
deliberations. As a historical question, explaining why a WG made a
particular decision is never easy.)

Michael Kay
http://www.saxonica.com/

Current Thread