[class.union.anon] Correct definition of union-like class

[TymianekPL]

There is a redundant union or, as all unions are classed.

-  union-like class is a union or a class [...]
+  union-like class is a class [...]

[eisenwave]

I don't think there's something objectively wrong here. "union" is a defined term in https://eel.is/c++draft/class.union.general#def:union

Saying "union or class" is clunky though. My preference is "class type", if we want a change, but just "class" is also fine by me.

[TymianekPL]

I suppose a synonym term for "Correct" could be used? But I'm not sure which one reflects the intent better...

[Halalaluyafail3]

I thought it was meant to be read like "A union-like class is a union or (a class that has an anonymous union as a direct member)." rather than "A union-like class is (a union or a class) that has an anonymous union as a direct member.".

[TymianekPL]

Okay turns out that union-like class is just a leftover from old wording that doesn't exist anymore

Should we remove this definition as it's unused?

[eisenwave]

I thought it was meant to be read like "A union-like class is a union or (a class that has an anonymous union as a direct member)." rather than "A union-like class is (a union or a class) that has an anonymous union as a direct member.".

Oh yeah, that reading makes much more sense.

Perhaps it could be clarified by "is a union or is a class"

[eisenwave]

Okay turns out that union-like class is just a leftover from old wording that doesn't exist anymore

Should we remove this definition as it's unused?

At the end of the paragraph, it says

At most one variant member of a union may have a default
member initializer.

So the "union-like class" term may not be used elsewhere, but it's used internally.

[TymianekPL]

Oh right true, interesting...
I'm still unsure about how one should parse this sentence still. Maybe we can come up with better wording to say this intent? The current one is a bit ambiguous 😅