[enhancement] deprecate Cardinality _MANY

[reinhapa]

Description:

Deprecate _MANY as it is not an available XPath indicator.

Use EMPTY_SEQUENCE and EXACTLY_ONE to achieve the same behavior if needed.

[adamretter]

@reinhapa I think your code comment could benefit from being a little less assertive as _MANY has the value 4 whilst ONE_OR_MORE has the value 6. So it's unfortunately in most places not just a case of replacing one with the other, but instead at those call sites some changes may need to be made

[line-o]

_MANY will be removed without an (exposed) replacement. In places where _MANY is still used within the codebase it needs to be adapted to make use of ONE_OR_MORE instead.

I think the comment is good enough to transport that information

[adamretter]

@line-o actually no, that's not how you would fix the call sites, which is why I commented that Patrick's comment in the code is incorrect/misleading

[reinhapa]

@reinhapa I think your code comment could benefit from being a little less assertive as _MANY has the value 4 whilst ONE_OR_MORE has the value 6. So it's unfortunately in most places not just a case of replacing one with the other, but instead at those call sites some changes may need to be made

@adamretter What would be your suggestion instead?

I had a look into the code and to occurrences of those values and for me at least there is a mix up between the following use cases

• XML schema definition: ZERO_OR_ONE, ONE_OR_MORE, ZERO_OR_MORE)
• Function arguments: EXACTLY_ONE, ZERO_OR_MORE
• Functions results: EMTY_SEQUENCE. EXATLY_ONE, ZERO_OR_MORE

IMHO this can not actually be captured in a single Enum in general.

[adamretter]

I had a look into the code and to occurrences of those values and for me at least there is a mix up between the following use cases

XML schema definition: ZERO_OR_ONE, ONE_OR_MORE, ZERO_OR_MORE)
Function arguments: EXACTLY_ONE, ZERO_OR_MORE
Functions results: EMTY_SEQUENCE. EXATLY_ONE, ZERO_OR_MORE

The cardinality Enum should only have the cardinalities applicable to XDM, i.e.:

• EMTY_SEQUENCE ()
• ZERO_OR_ONE (?)
• EXACTLY_ONE
• ONE_OR_MORE (+)
• ZERO_OR_MORE (*)

My suggestion is to fix the calling code by refactoring it, the solution to fixing that is not to move those call sites to ONE_OR_MORE as that will just break the current behaviour.

[reinhapa]

My suggestion is to fix the calling code by refactoring it, the solution to fixing that is not to move those call sites to ONE_OR_MORE as that will just break the current behaviour.

That is actually the plan. But I would suggest to move the discussion to #6007 ...