Joe Kesselman wrote:
>...
>MOVED:
>3) And that we send a formal request to the XSLT Working Group that
>this concept be nailed down more explicitly in XSLT 2.0.
+1 to that.
In accordance with my other posting on this thread, we could request
that "extension attributes" are allowed as a means to choose among
conforming variants of the results, where there does not already
exist (in the XSLT specs) a method to choose the variant. There
would be some rules about permissible namespace URIs, etc. Do we
want such attribute values to be static-only (no AVTs)?
Conceptual test: If, given a vanilla (no extensions) stylesheet and
an unidentified XSLT processor, the processor would be allowed to
produce either Variant A or Variant B, then it is permissible for
the processor to accept "extension attributes" to cause either
Variant A or B to be produced deterministically. I don't like the
idea that randomness is allowed, and the spec seems to oppose all
unpredictable behaviors.
.................David Marston
