Stream: t-compiler/wg-nll

Topic: issue-51348-ICE


nikomatsakis (Jun 27 2018 at 17:05, on Zulip):

So @Jon Gjengset I can certainly reproduce #51348...

Jon Gjengset (Jun 27 2018 at 19:08, on Zulip):

I implemented a workaround for the issue, but now I hit another ICE: https://github.com/rust-lang/rust/issues/51649

nikomatsakis (Jun 27 2018 at 19:12, on Zulip):

heh, ok, that's another known ICE yes

lqd (Jun 27 2018 at 19:12, on Zulip):

(I think this is the one hitting bootstrap — and maybe futures but haven't been able to look into that one yet)

nikomatsakis (Jun 27 2018 at 19:14, on Zulip):

yeah

nikomatsakis (Jun 27 2018 at 19:14, on Zulip):

I know the problem

nikomatsakis (Jun 27 2018 at 19:14, on Zulip):

I can work on a fix

nikomatsakis (Jun 27 2018 at 19:14, on Zulip):

as soon as I satisfy @Eduard-Mihai Burtescu's many nitpicks :)

nikomatsakis (Jun 27 2018 at 19:14, on Zulip):

kind of have to land the existing PR to fix that one

Jon Gjengset (Jun 27 2018 at 19:22, on Zulip):

It looks like that other ICE is even more nefarious. I can reproduce even with a trivial Deref impl: https://github.com/rust-lang/rust/issues/51649#issuecomment-400799282

nikomatsakis (Jun 28 2018 at 10:33, on Zulip):

ok so the main PR landed

nikomatsakis (Jun 28 2018 at 10:33, on Zulip):

I will try to fix the "other ICE"

nikomatsakis (Jun 28 2018 at 10:33, on Zulip):

but I guess I'm in the wrong thread now =)

nikomatsakis (Jun 28 2018 at 12:38, on Zulip):

@Jon Gjengset hmm I cannot reproduce your ICE now... but I fear something is still going wrong, just not ICEing anymore

nikomatsakis (Jun 28 2018 at 12:38, on Zulip):

(it does ICE with nightly)

Jon Gjengset (Jun 28 2018 at 14:44, on Zulip):

This is the one that fixes 51348? Great! :D Though now with my workaround it doesn't terribly matter. Only the "other" ICE does.

nikomatsakis (Jun 28 2018 at 14:45, on Zulip):

I forget which ICE is which :)

Jon Gjengset (Jun 28 2018 at 14:45, on Zulip):

As for not reproducing on master, I can try today's nightly and see if my full use-case triggers it

Jon Gjengset (Jun 28 2018 at 14:45, on Zulip):

51348 is the multi-match with guard

nikomatsakis (Jun 28 2018 at 14:45, on Zulip):

ah, I see, no I was talking about #51649

Last update: Nov 22 2019 at 01:00UTC