Some issues in next branch

[nkutsche]

Hi @dmj,

I've found some issues in the test cases of the next branch. Most are just careless mistakes (I hope), so I don't think it is worth to make for each a separete ticket:

1. rule-based-validation/let-value-as.xml has the wrong default phase
2. pattern-templating/*-abstract-name-path.xml (1, 2) produces something like <sch:name path="name()"/> instead of <sch:name path="."/>
3. instance-document-selection/pattern-subordinate-document-from.xml has <sch:phase name="">, should be <sch:phase id="".
4. reporting/svrl-*-lang.xml (1, 2, 3, 4): each contains an assertion like //svrl:failed-assert[lang('de')]. Correct would be //svrl:failed-assert/svrl:text[lang('de')], wouldn't it?
5. reporting/svrl-name-path.xml: the assertion normalize-space(//svrl:successful-report) = 'attribute' may correct, but it would be more precise with normalize-space(//svrl:successful-report/svrl:text) = 'attribute'.
6. schema-composition/schema-phase-when.xml: two things...
   • the when should be /root/element instead of element, shouldn't it?
   • the test case has a generall problem: there is no difference if the when triggers the phase or the default behavior (#ALL) is applied. I would recommend to make the when-phase valid and introduce a second phase that would make the result invalid. The expected result should be valid then.

Do you prefer a pull request with my fixes?

[dmj]

Thanks for working with the conformance tests!

1. -- This is an error, yes.

2. -- Not an error.

The specifications says:

The optional path attribute is an expression evaluated in the current context that returns a string that is the name of a node.

(I think the choice of naming the attribute path is odd, but the conformance test is correct here.)

3. -- This is an error, yes.

4. -- I think it is correct this way.

The @xml:lang attribute sets the language property of an element and specifies the language used in the contents and attribute values (see XML 5th Edition, 2.12).

The langue property of a sch:rule is inherited by the sch:assert element. If we check that there is a svrl:failed-assert element with the proper language property, we not just check the language property for the assertion message, but also for the attributes @test, @role, @flag.

5. -- Not neccessarily.

svrl:text as child of svrl:failed-assert was introduced in the 3rd edition of ISO Schematron in 2020. Before that, the assertion message was not wrapped.

6. -- This is an error, yes.

I pushed fixes for the errors. It was easy to check, confirm & fix.

[nkutsche]

The specifications says:

The optional path attribute is an expression evaluated in the current context that returns a string that is the name of a node.

(I think the choice of naming the attribute path is odd, but the conformance test is correct here.)

That is really suprising. Not only the name of the attribute is odd here. What shall be the difference between <sch:value-of select="name(foo)"/> and <sch:name path="name(foo)"/>?? And what if I make something like <sch:name path="string(foo)"/>? It does not make any sence to me. And it is not suprising to me, that e.g. Oxygen does it implement in that way how I understood it.

The @xml:lang attribute sets the language property of an element and specifies the language used in the contents and attribute values (see XML 5th Edition, 2.12).

The langue property of a sch:rule is inherited by the sch:assert element. If we check that there is a svrl:failed-assert element with the proper language property, we not just check the language property for the assertion message, but also for the attributes @test, @role, @flag.

Hm, but according to the SVRL schema there is no xml:lang attribute allowed on svrl:failed-assert.

svrl:text as child of svrl:failed-assert was introduced in the 3rd edition of ISO Schematron in 2020. Before that, the assertion message was not wrapped.

Ok, that make sence.

[dmj]

Hm, but according to the SVRL schema there is no xml:lang attribute allowed on svrl:failed-assert.

True enough. And this is a problem: Schematron/schematron-enhancement-proposals#89 -- @role, @flag, @test lose their language property.