Make PropertyNamingStrategy not affect Enums

[JooHyukKim]

possible fix for #4409

[cowtowncoder]

There are a few issues here; I don't think this would work.

First question is compatibility: do we know when did PropertyNamingStrategy start applying to Enums (I know it's indirect and not explicit change, i.e. side effect)? If it was 2.16 we could I guess consider 2.16 patch -- but if 2.15, it definitely would need to wait until 2.17.

Second: determination for renaming should be based on type being introspected; PR does checks on per-property basis but it should really be based on type of POJO that POJOPropertiesCollector is introspecting. So check needs to be earlier before loop.
But it looks like there are some concerns (failing/regressed test I assume) wrt @JsonFormat(shape = Shape.OBJECT) applied on Enum type, or perhaps even property.
Not 100% sure how that is to be tackled, but it cannot really be checked from within place check is now.

[cowtowncoder]

This is in wrong place: it would check for annotation of a property of Enum type referring to another type and being annotated as Shape.OBJECT -- that would not cover cases where Enum class is annotated or -- if supported, not sure it is -- property of another (almost certainly non-Enum) type pointing Enum type and having Shape override.

[cowtowncoder]

How about this: a more extensive change to block any application of global PropertyNamingStrategy for enums; adding test with specific Enum entries that would be affected -- and merging that test in 2.15 to show before/after change?

[JooHyukKim]

How about this: a more extensive change to block any application of global PropertyNamingStrategy for enums;

Makes sense 👍🏼 Could you share with me if you already have any idea on how to approach things?

adding test with specific Enum entries that would be affected -- and merging that test in 2.15 to show before/after change?

You mean like test I wrote in PR against 2.15 in my personal repo? I can't think of other cases at the moment, but will figure out.

Btw, should we keep the work against 2.16 then?
cc. @cowtowncoder

[JooHyukKim]

Implemented 20b6e54 as per comment #4403 /cc @cowtowncoder

[cowtowncoder]

LGTM, will try to get merged later tonight.