resultMap.hasResultMapsUsingConstructorCollection should stay false if type handler is specified

[harawata]

Hi @epochcoder ,

This is a follow up of #3108 .
If typeHandler is specified for a ResultMapping, it should not enable hasResultMapsUsingConstructorCollection, I think.
Could you verify if this assumption is valid?

I thought you would do this to fix CollectionInConstructorTest.testCollectionArgWithTypeHandler().
How did you fix that? 🤔

[coveralls]

coverage: 87.27% (-0.02%) from 87.289%
when pulling d1238a6 on harawata:3108-follow-up-type-handler
into 2de2506 on mybatis:master.

[epochcoder]

Hey @harawata, in this case the mentioned test was fixed by allowing pending creations for simple result maps. (which also fixed some other cases). I cannot think of a reason (yet) why a type handler presence should disable it. If a collection is present in a constructor, we want to defer until later, whether that collection comes from a type handler or not.

What I think we should do to validate this is try and think of a (test) case where this does not hold true.

For instance if we have a mapping with a type handler that creates a list, and a nested result mapping, we still want the behavior.

Maybe we can try and come up with a constructor and mapping as a test case that could prove the doubt? :)

[harawata]

If a type handler is specified, it means that the collection is created from (a single column of) a single row, so there is no reason to defer the creation.
In the case of aforementioned test, the collection is created from a comma separated string.
This was actually the only way to use collection as a constructor argument before #3108 .

[epochcoder]

Ahh, I think I understand, you basically just want to ignore the pending creations if there is a type handler specified, however if any other mappings have a normal collection it will get enabled again (due to the for loop).

in this case I think it is okay, but I would still like to add another test case to this PR then, that includes such a mixed mapping :)

[harawata]

I was trying to edit my comment, but that is exactly right. :D

Regarding the test case, I noticed this because one of tests I was writing for #556 failed.
I'll push it to #3388 soon (maybe tonight).

[harawata]

I would still like to add another test case to this PR then, that includes such a mixed mapping

It is not the test I mentioned, but this is definitely a good idea.
If you can add one to this PR, please do.

[harawata]

Thank you, @epochcoder !