@nikomatsakis or @pnkfelix: I'm looking at this comment, the first linked test was tackled by a PR of mine a while ago. The second linked test is missing an error - is that correct or should I look into it? (current AST/current NLL)
it's not obvious why the AST checker gets an error there to me
looking a bit more
oh, I .. guess I maybe see
looks kind of like a bug to me
In NLL? We should be producing that error?
right, in NLL
feels like when we invoke
iter, we have to infer the lifetime of the trait to be
I think we should file a NLL-sound bug on this
and assign it to you, @davidtwco :)
no good deed and all that
(I would be happy to help investigate; I think the first thing is we should check out the region constraints that get generaed)
Unless I've made a mistake, I think the error does come up after the type mismatch is fixed.
ah, that..is plausible
we really should work on making rustc less "phase-y"
probably the type mismatch can be removed
So there isn't an issue here, other than the test being strange?
I don't think it's imp't to the test per se?
yeah, I guess not
I can't see why it would be.
I guess I don't know what that is testing
(easier than browsing for that)
For the input file below, the invocation of rustc that prompts with the "consider using an explicit lifetime parameter" hint is providing an erroneous signature.
ok, seems like fixing the type mismatch error is "truer" to the intention of the test
I'll submit a small PR for that.