[class.cdtor] Only objects of scalar type can be accessed

[languagelawyer]

No description provided.

[jensmaurer]

I think that's right, but @zygoloid should also comment.

[zygoloid]

I think this is correct, but I think we're missing similar wording talking about accessing the object representation of the object.

The old wording -- in particular, the "the object or" part -- might have been read as covering that before, in some cases: an incautious reader might conclude that copying a union with a trivial copy special member results in an access of the value of the union object.

I think we want "the value of any of the object or its subobjects or any element of its object representation" in the first edit and "the value of the object or subobject thus obtained" in the second edit, though that seems to be straying out of editorial country.

[tkoeppe]

Thanks, @zygoloid! Let's give this some more baking time, not for now.

[tkoeppe]

@languagelawyer: please see above discussion, and also rebase?

[languagelawyer]

@tkoeppe If I understand it correctly, @zygoloid suggests that this can't be fixed editorially.

[tkoeppe]

@languagelawyer: could you make the proposed change regardless, and we can run this issue past CWG? If necessary we can file a core issue then, but maybe a group review will suffice.

[jensmaurer]

Put on CWG's agenda: https://wiki.edg.com/bin/view/Wg21telecons2021/CoreWorkingGroup

[AlisdairM]

I could not find any minutes --- were the editorial issues flagged for telecon ever discussed?

[tkoeppe]

@jensmaurer Just to check, is the current state either self-evidently editorial, or did CWG otherwise have a look?

[jensmaurer]

The change looks correct to me. CWG did not have a look.

[tkoeppe]

OK, let's merge it then, thanks!