Stream: t-compiler

Topic: Universes

Matthew Jasper (Jan 06 2019 at 12:37, on Zulip):

@nikomatsakis What's the expected behaviour for the examples here:

nikomatsakis (Jan 07 2019 at 15:47, on Zulip):

@Matthew Jasper I expect errors for both — I'm not sure why the second one works with NLL, that's curious

nikomatsakis (Jan 07 2019 at 15:49, on Zulip):

the reason that I expect an error from the second one:

the T in the impl is bound outside of the fn, so it can't name regions bound on the fn type, but we are trying to equate it with for<'a> fn(&'a u8).

Reduced: exists<T> { forall<'a> { fn(T) = fn(&'a u8) } }

Matthew Jasper (Jan 15 2019 at 20:47, on Zulip):

Finally raised as #57642

lqd (Jan 15 2019 at 20:49, on Zulip):

(I updated the link to not 404)

lqd (Jan 15 2019 at 20:52, on Zulip):

@Matthew Jasper do you think the unclear message part of this is a duplicate of #57362 or just similar ?

Matthew Jasper (Jan 15 2019 at 20:58, on Zulip):

It's pretty similar. They'll probably get solved by the same PR, so long as the PR takes a general approach

nikomatsakis (Jan 16 2019 at 14:08, on Zulip):

thanks @Matthew Jasper

Last update: Jul 14 2020 at 13:45UTC