These words are quick to anger... For @zygoloid to decide whether this is editorial or needs CWG review.

I'm confident that the old and new words have the same meaning, but I'd like CWG review to double-check that we don't intend for these two cases to be handled differently any more. (As I recall, they did actually do different things before we reformulated this in terms of implicit conversion sequences in CWG1895.)

resolved a conflict with cf283d8

@jensmaurer, @zygoloid: Did this PR ever come up in CWG?

I suspect that CWG2316 is related, though I can't see the detailed description.

@cubbimew Rebase this regardless please if you don't mind, so that CWG can look at current wording.

@cubbimew: rebase please?

rebased, sorry for delay (was in conflict against beb8815 )

Approved by CWG telecon 2020-08-17.

Not that anyone implemented it that I know of, but this reverts the resolution to CWG 2321 (http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2018/p1113r0.html) adopted by the June 2018 Rapperswil meeting.

Compiler Explorer link demonstrating (at this time) non-adoption of the resolution to CWG 2321: https://godbolt.org/z/f8KqaM.

I don't understand, now, how that resolution was necessary anyway: if you fall through to the other bullet in the issue's example, isn't the target type A, to which const B can be implicitly converted?

The change was reverted in 0247412 because the change reverts the normative effects of CWG 2321. Such action is not editorial, and furthermore unintended.

I don't understand, now, how that resolution was necessary anyway: if you fall through to the other bullet in the issue's example, isn't the target type A, to which const B can be implicitly converted?

Maybe that's why MSVC accepts but with different cv-qualification for the result than CWG 2321 would give.