Re: Carmelo's Post on Gathering Discretionary Implementation Issues

[Tony Graham <tgraham@mulberrytech.com>]

At 27 Jun 2000 15:24 -0400, David_Marston@lotus.com wrote:
 > Tony has a great summary of the items we need for each identifiable
 > discretionary provision in the spec. I've already remarked about the

I'm glad that you like it.

 > "XSLT Recommendation section reference" item. Here's one that we

Yes, we do need more accurate ways to refer to portions of the
Recommendations.

 > should review:
 > 
 > >A question, for use with the eventual "rendition control instance",
 > >that can be answered either yes, signal the error, or no, perform
 > >the alternate behaviour.
 > 
 > Are we sure that all the discretionary items are boolean? There may

I was only working on the discretionary error recovery portions, and
at this point I'm not sure that they, at least, are not all boolean.

I don't know that all discretionary items are boolean, but it would
make the "rendition control instance" (and its user interface) simpler
if all items are boolean.

 > also be other things we could express as questions, where the ease of
 > answering the question would be beneficial for acceptance of this
 > work.

I agree about the benefit of asking questions (hopefully simple
questions) with simple answers.

 >       To take one of the current items of discussion as an example:
 > if we ask the vendor/implementor to describe their naming scheme
 > for generated namespace prefixes, that helps in predicting the exact
 > output that would be correct *for that processor* on a test case that
 > has generated namespace prefixes.

In terms of conformance, I think that checking the namespace prefixes
is a non-event since there is no requirement (and should not be a
requirement) to use the same prefixes as in the "name" attribute.

However, if checking the test output against a known-good result file
is done by simply comparing the two files, e.g. by using diff, then
arbitrary prefixes will create problems.  Knowing the format of the
prefixes may ease the problems if, as we discussed, the test and
result files are generated anew for each processor configuration.

Actually, checking namespaces in test results will be interesting
since the the processor has some options about where in the output it
declares the namespaces; e.g. all on the document element or on the
top-most element or even all elements "in" a particular namespace.

You could write a stylesheet to check the namespace usage in a test
result, but does that require a "known-good" XSLT processor in
addition to the one being tested?

Regards,


Tony Graham
======================================================================
Tony Graham                            mailto:tgraham@mulberrytech.com
Mulberry Technologies, Inc.                http://www.mulberrytech.com
17 West Jefferson Street                    Direct Phone: 301/315-9632
Suite 207                                          Phone: 301/315-9631
Rockville, MD  20850                                 Fax: 301/315-8285
----------------------------------------------------------------------
  Mulberry Technologies: A Consultancy Specializing in SGML and XML
======================================================================