Dave Abrahams
Dave Abrahams
I suggest reducing the test case first if possible. What architecture is your machine?
Might it be better if name lookup did not ignore declarations with errors?
That's not what I'd expect. I'd expect that it _does_ have a member `foo` but that `foo`'s type is indeterminate (usually represented by some internal error type).
> I am not convinced that would solve the problem you are bringing up Nobody said it would solve the problem. I only said it could help us solve it....
> we DO require contributors to run swift-format for every PR. Yes. You also require them to run the swift compiler, but you don't make them download the sources and...
SInce there's been no counterargument, I'm probably going to remove this dependency soon unless someone raises an objection here.
This will become a non-issue to the extent that we stop using SPM. However, the documentation will still need to be updated. See #1300
@lucteo let's try to move open PRs forward and get them merged, or close them please.
@adevress Sorry for the slow response time. I think I'm going to move forward with #1318; shall we discuss how the docs should be changed?