Issues with XML test cases

[chrdebru]

• No test case for testing selecting values from attributes see A test case for selecting values from attributes (XML only) #107
• Should we have a test case for selecting elements based on conditions?
• More test cases for language maps? E.g., ensuring no terms are generated when the language map does not yield terms..

[DylanVanAssche]

• No test case for testing selecting values from attributes: +1, feel free to add it if you have one.
• Should we have a test case for selecting elements based on conditions?: Tricky XPath is welcome I suppose? People do tricky stuff with it so we should check it?
• More test cases for language maps? E.g., ensuring no terms are generated when the language map does not yield terms.: For now we have the basics, but there might be corner cases left open, so all help is welcome. This is mostly for the maintainers to figure out?

[chrdebru]

I agree. I can add a test case for the first if @pmaria and @andimou agree that selecting attributes must use valid XPath queries.

[chrdebru]

@DylanVanAssche : The first should be solved with #107.

[dachafra]

I agree with @chrdebru on both pending tasks.

I also have many use cases with very tricky XPath expressions, that currently are not supported by many engines.

My question is: do we need them now? If yes, who wants to create them? If not, I'll mark it as working-group @DylanVanAssche @chrdebru

[pmaria]

• Should we have a test case for selecting elements based on conditions?

I also have many use cases with very tricky XPath expressions, that currently are not supported by many engines.

These would fit best in the io-registry IMHO.

[DylanVanAssche]

I agree with @pmaria Tricky XPath expressions need to be specified and tested in rml-io-registry.
We need to collect them first to have a good idea of what is called 'tricky' as that's very vague currently.

[dachafra]

More test cases for language maps? E.g., ensuring no terms are generated when the language map does not yield terms..

and what about the second? Do we need it? @DylanVanAssche

[DylanVanAssche]

and what about the second? Do we need it?

Depends if the spec requires that if no language tag is generated for the triple while there's a language map, that we don't generate a triple --> basic reference algorithm + spec definition is needed.