[2018-11 CWG Motion 15] P1141R2 Yet another approach for constrained declarations

[jensmaurer]

P1141R2

[tkoeppe]

Note the interaction with motion CWG-4 (#2397). We'll need to adjust the wording of that paper after we apply this one.

[burblebee]

I'll create the branch for this motion based on the paper as is, then I/someone can add the required changes as a separate edit, OK? @tkoeppe Can you suggest the needed changes (or just apply them to the branch yourself after I commit)?

[tkoeppe]

Right, the fixup should probably appear in a separate commit (but in the same branch).

The point of contention is the grammar production qualified-concept-name that we are deleting in this paper, but which is being used by the earlier paper. We need to rephrase the affected sentence.

[hubert-reinterpretcast]

-> type-constraint

If the return-type-requirement is of the form -> type-constraint, then the type-constraint applies to decltype((E)). The immediately-declared constraint of decltype((E)) shall be satisfied. [Note: ...]

[burblebee]

@hubert-reinterpretcast Your suggested change looks correct, but I'm not sure about the wording, in particular "the type-constraint applies to decltype((E))" - can we find better wording than "applies" here? Perhaps wording that is closer to the original text?

[hubert-reinterpretcast]

In the language of P1141:

-> type-constraint

If the return-type-requirement is of the form -> type-constraint, then the contextually-determined type being constrained is decltype((E)). The immediately-declared constraint of decltype((E)) shall be satisfied. [Note: ...]

[burblebee]

@hubert-reinterpretcast Much better, thank you!

The wording changes were added to motions-2018-11-cwg-4 since it had the wording which needed to be changed.