[PATCH] D30776: [coroutines] Fix diagnostics depending on the first coroutine statement.

Eric Fiselier via Phabricator via cfe-commits cfe-commits at lists.llvm.org
Fri Mar 10 15:28:09 PST 2017


EricWF added a comment.

In https://reviews.llvm.org/D30776#697258, @GorNishanov wrote:

> In https://reviews.llvm.org/D30776#697233, @EricWF wrote:
>
> > Good to know. I'll update this tomorrow.
>
>
> Well, that is just the thought. Possibly we can check for the types of await_ready and await_suspend in BuildResolvedCoawaitExpr. Then, we don't have to keep the vector of await/yield-expressions.


I guess it depends on when we want the diagnostics to be issued. And I think emitting them as they are built makes the most sense. If we attempt to issue the diagnostics at the end of the body it may be too late.

Do you still want me to put the vector back in now or when the need actually arises?


https://reviews.llvm.org/D30776





More information about the cfe-commits mailing list