[www-qa] W3C Recommendations should present discretion in an organized form

Presented for your edification: a document used by the OASIS Technical
Committee (TC) on XSLT/XPath Conformance to organize information about
discretionary choices in the behavior of an XSLT processor. Nearly all
aspects of processor behavior are defined by a single standard, yet
there are about 50 different ways in which a processor can vary while
still conforming to the W3C Recommendation.

Our TC is using this in cataloging test cases, because some tests only
apply to those processors that made a particular design choice. You may
find other value in this document depending on your point of view:
+ Processor developers can see what design discretion they have
+ Tool developers embedding a processor can see the variations that
  affect how they invoke the processor, or even which one they embed
+ Test labs can see what they need to know about a processor before
  subjecting it to testing
+ Web site developers have more checklist items to consider when making
  decisions about XSLT technology
+ Documentation writers can see possible new content requirements.

That last item may also be the main area of interest to the QA Activity.
For example, there could be a documentation standard insisting that
every Recommendation have an appendix that collects all discretionary
choices in one place. When you look at this document and see the messy
XPointer-style references into the Rec, you may wish that all Recs had a
better set of anchors/markers/IDs within themselves. I find it reasonable
to designate an XML document as the normative foundation, even when the
XML is not a linear presentation, and this document is an example because
it was derived by XSLT transformation from a non-linear XML file. (The
questionnaire for XSLT developers will be derived from the same XML, as
will certain entries in the catalog of test cases.)

Undoubtedly, further refinements will occur to this document. For now,
it has reached a plateau of maturity that warrants showing it to the
WWW-QA community for expanded discussion.
................David Marston

(See attached file: newdisc.html)