@blitzerr I was thinking this would be a good "next PR"
I suspect it's as simple as supressing the message if the variable in question is not marked as "user-defined"
Sure, I can take a look at that.
Varkor commented that his his pr will fix the issue. You are not convinced ?
I am not convinced
I can check though; I'm not sure what @varkor meant by that actually
I think they may have meant that they will be giving more errors earlier on but maybe I misunderstood them
oh, I am totally wrong :)
@varkor did include a fix, though it seems only semi-related to the rest of the PR
that said, the fix looks odd to me :)
but we should find another issue I guess
left a comment; I also think we could make this particular change earlier
" I also think we could make this particular change earlier"
I didn't get that part
sorry, only just seen these messages
to summarise: I added a fix for that issue after it appeared as an issue in the uninhabitedness PR — it wasn't immediately related, but because at the time it hadn't been reported, I thought I'd quickly add a fix then and there
I was also in a bit of a rush, so my "solution" was a bit hacky
I'm testing an updated version now, which seems more like the correct fix
didn't realise someone else wanted to look at this, sorry!
@varkor do you plan to open a separate PR with the fix?
it depends whether the uninhabited PR seems ready now / very soon — then it'd be slightly less work for bors
I'm happy to split it out into a separate PR if that's clearer, though
(I can do that tomorrow if so)
I've split out the fix into https://github.com/rust-lang/rust/pull/54787