test: verify non-existing function invocation

[nikku]

Yields null.

---

Closes https://github.com/dmn-tck/tck/issues/539

[StrayAlien]

looks good to me @nikku . Thanks. I really thought we'd have this covered already - even accidentally - but, I don't think we do. Good catch. Though, just a suggestion, perhaps we could take the opportunity to examine some other simple related cases here?

Say ...

• null() (returns null)
• "somefunc"() (returns null), where a function called somefunc is in scope
• and similar to the above, invoking non functions as though they were functions. The grammar says function invocation is function invocation = expression , parameters where expression can be (well) any expression including literals. So perhaps we could exercise a few, like 123(), @"2023-11-11"(). Perhaps even a double invocation likefunc()()wherefunc` exists in scope as a function.

Happy for comments.

[nikku]

@StrayAlien I'll add the cases you mentioned.

[nikku]

@StrayAlien checkout https://github.com/dmn-tck/tck/pull/612/commits/072ba7b733b90c8738ba66fb03ab0497ca1e656e which adds the additional test cases.

[StrayAlien]

Thanks @nikku . I know they might seem a little (perhaps) silly, but, we're here to catch as many angles as we can. Maybe a new runtime would not actually do 123(), who knows really. Perhaps the mixup of literal vs identifier for the "some_func"() thingo might catch people, but, hey, I wish I has a € for every time some silly test saved my *ss. :-) Easy to pass tests are 'free', but they're still tests ... 👍

[nikku]

@StrayAlien There is nothing better than testing different (edge) cases. So no, testing such cases does not sound silly at all.

The DMN TCK is a good case of I-Don't-have-to-read-the-freaking-manual. Which is large :wink:. One more reason to support this.