Clarify support for XPath 4.0 or XSLT/XQuery 4.0

[galtm]

Should I be able to use XPath 4.0 expressions in a Schematron schema, if the underlying processor supports them? How about XSLT 4.0 or XQuery 4.0? Would I use queryLanguageBinding="xslt4" or queryLanguageBinding="xquery4" on the schema?

The XSpec repo has a feature request issue about supporting 4.0 in XSpec tests, but I didn't know if supporting 4.0 in XSpec tests for Schematron schemas was supposed to be part of that task.

[AndrewSales]

The query language bindings are defined in normative annexes to the standard (I've been working on the XQuery ones just today).

I think while the 4.0 specs are still in draft, we would have a hard time referencing them normatively in the Schematron standard. So I would say if implementers choose to implement 4.0 bindings at this juncture, that is a matter for them.

But thank you @galtm for the reminder to reserve those binding values for future use, which I have now done in the latest working draft.

I'll leave this open for now, in case circumstances change and while further technical changes to the draft are permitted.

[galtm]

Thanks, @AndrewSales, for your reply and for reserving the 4.0 binding value for future use.

As for the XSpec backdrop to my question, there is a way to test a Schematron schema that uses 4.0, but we can consider it experimental for now.

[rjelliffe]

First off, you can use any query language binding your implementation provides, no matter whether they are in the standard or not. The standard defines a set of standard query language bindings: so that e.g. if you select "xslt2" as your QLB your implementation should provide a QLB with whatever the standard specifies for that QLB (with the obvious and usual caveat that where a standard QLB is vague or incomplete, the implementation should do something sensible and useful, preferably in concert with other implementators, preferably documented, and possibly with a view to getting the gap in the standard filled-in.) However, that in no way limits what non-standard or pre-standardization QLBs that a Schematron schema can use. (I think most people would be comfortable with the rule of thumb that experience is the best foundation for a standard QLB, not speculation.) An implementation could allow Java functions as a QLB for example. However, the problem here is that xpath4 does not actually exist yet, as far as I know. So it is not possible for an ISO standard to reference something "normatively" before W3C (or whoever) has said what that thing specifically is. What the ISO standard can do is reserve the names "xslt4" and "xpath4" (if it does not already)/ ISO could possibly give a non-normative annex that gives the best guess of the binding with the caveat that it might be superseded with a name like "preview-xslt4" or whatever. (In particular, upgrading the Xpath to Xpath4 is trivial; the thing that can be tricky is whether XSLT4 has elements or attributes that an "xslt4" QLB should demand. This is like how Schematron needed an ***@***.*** to cope well with XSLT3 and probably XSLT2.) So I suggest that anyone who is implementing changes to a Schematron engine to make use of the "xpath4" features currently under discussion, or that are in experimental SAXON implementations, should use a QLB name like "draft-xpath4". That makes it's status clear. After XPath4 is baked at W3C and available, then the implementer would be safe to provide "xpath4" or "xslt4" with just the xpath4 features because any standard QLB will have those; but I think if implementations allow "xslt4" to access non-Xpath4 XSLT4 features before the ISO QLB is standardized, they would be mislabelling the QLB (which kinda causes the problem that having a QLB name is supposed to prevent.) If a user puts "xslt4" in the QLB, and an implementation does not support the ISO standard QLB (even a late draft) because the ISO QLB not baked, then the implementation should tell the user "no, that is not a real thing" to prevent the chances of false positives or negatives. Validation is one of those things where you really do prefer certainty and strictness about this kind of thing as the default operational mode of the validator, yes? That is my idea of the way things should work, anyway. Regards Rick

…

On Mon, Mar 18, 2024 at 2:21 AM Amanda Galtman ***@***.***> wrote: Should I be able to use XPath 4.0 expressions in a Schematron schema, if the underlying processor supports them? How about XSLT 4.0 or XQuery 4.0? Would I use queryLanguageBinding="xslt4" or queryLanguageBinding="xquery4" on the schema? The XSpec repo has a feature request issue <xspec/xspec#1729> about supporting 4.0 in XSpec tests, but I didn't know if supporting 4.0 in XSpec tests for Schematron schemas was supposed to be part of that task. — Reply to this email directly, view it on GitHub <#70>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AF65KKM2ULQTPUBHYIOBSH3YYWYIBAVCNFSM6AAAAABE2JLZEOVHI2DSMVQWIX3LMV43ASLTON2WKOZSGE4TANZSGAYTCNY> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>