[class.mem.general] Allow non-defining declarations of nested classes. #4416
Conversation
|
OK, but I'm not 100% sure this is editorial. Could we get CWG sign-off on this, or at least a second pair of core eyes? |
|
@zygoloid , could you please take a look? |
There was a problem hiding this comment.
I have no doubt that this change reflects the CWG intent, and that the wording and example in [class.nest]p3 are correct.
As for whether that's enough to apply this editorially without explicit CWG oversight... perhaps it's worth sending an email past the core reflector to check if anyone would prefer this to be handled non-editorially.
| The \grammarterm{member-declarator-list} can be omitted only | ||
| after |
There was a problem hiding this comment.
I think "can be omitted only after" is problematic phrasing -- it's unclear whether it's giving permission, stating a restriction, or both. (Can a member-declarator-list be omitted after any elaborated-type-specifier, or only some of them?) This is made a little worse by this change, because three of the four top-level grammar productions for elaborated-type-specifier aren't allowed in this context, per the end of [dcl.type.elab]p1. This might help:
The \grammarterm{member-declarator-list} shall not be omitted unless
the declaration is a friend declaration or its sole constituent is
- ...
|
Thanks a lot! That seems clear, let's put this on the CWG list and get sign-off. |
Fixes #4409