Additional discretionary items in XSLT spec

[David_Marston@lotus.com]

Tony provided a convenient list of the discretionary provisions that relate
to error handling. Below, I provide additional items I found that
illustrate the breadth of discretionary items. The purpose of adding to the
list at this time is to help us think about a good naming scheme, a good
way to express the questions, and how we can obtain precise answers.

The only thing I've identified in XPath so far is the Note at the bottom of
Section 5.3: if there is an external DTD and the processor is not
validating, it still has the option to use the DTD to set defaulted
attributes.

3.3 Unparsed Entities
"...The XSLT processor may use the public identifier to generate a URI for
the entity instead of the URI specified in the system identifier...."

5.4 Applying Template Rules
The very last paragraph is a discussion of detection of infinite loops,
which we discussed informally in our Washington meeting. It's possible that
we could disregard this. To attempt to deal with it, we would get into
cataloging infinite-loop cases, then have a "do you detect this" question
for each such case.

7.1.2 Creating Elements
"XSLT processors may make use of the prefix of the QName specified in the
name attribute when selecting the prefix used for outputting the created
element as XML; however, they are not required to do so."  (Similar to what
Tony identified in 7.1.3.)

7.7.1 Number to String Conversion Attributes
The second of the two notes refers to language support. Should we just ask
if they support "en-US" for now, and not try to explore any other
languages? It says "NOTE: It is possible for two conforming XSLT processors
not to convert a number to exactly the same string. Some XSLT processors
may not support some languages. Furthermore, there may be variations
possible in the way conversions are performed for any particular language
that are not specifiable by the attributes on xsl:number. Future versions
of XSLT may provide additional attributes to provide control over these
variations. Implementations may also use implementation-specific namespaced
attributes on xsl:number for this."

10 Sorting
Has a note about language support parallel to the 7.7.1 note above.

11 Variables and Parameters
The end of the first paragraph refers to the ability to pass in parameter
values, and 11.4 says explicitly that the spec will not dictate how. My
reading is that a conforming processor does not have to offer any such
mechanism, but that would render xsl:param equivalent to xsl:variable. If
we want to cover that difference, we need to ask the vendor to specify the
ways in which parameters can be set externally.

12.1 Multiple Source Documents
"...The relative document order of two nodes in different documents is
determined by an implementation-dependent ordering of the documents
containing the two nodes. There are no constraints on how the
implementation orders documents..."

12.4 Miscellaneous Functions
Regarding generate-id(), it says "An implementation is free to generate an
identifier in any convenient way..." so we may want to ask for a
description of the way they're generated if we can't canonicalize it away.

16 Output
That first paragraph could be tricky: "An XSLT processor may output the
result tree as a sequence of bytes, although it is not required to be able
to do so (see [17 Conformance]). The xsl:output element allows stylesheet
authors to specify how they wish the result tree to be output. If an XSLT
processor outputs the result tree, it should do so as specified by the
xsl:output element; however, it is not required to do so."

16.1 XML Output Method
Bullet points about order of attributes and presence, placement (and
presumably order) of namespace nodes in the output. If we can canonicalize,
that should take care of all these concerns.

16.2 HTML Output Method
"The html output method may output a character using a character entity
reference, if one is defined for it in the version of HTML that the output
method is using."

17 Conformance
Probably not too devastating, except for its effect on attempts to run a
whole suite of tests: "A conforming XSLT processor may but need not recover
from any errors that it signals."

17 Conformance
"A conforming XSLT processor may impose limits on the processing resources
consumed by the processing of a stylesheet."