Skip to content

NLL ICEs from 3rd crater run #53482

Closed
Closed
@lqd

Description

@lqd

Following on the ICEs from #52217, here are the 591 occurrences of ICEs in the 3rd NLL crater run from #53426 (out of the 982 regressions).

Only 2 kinds of ICEs this time:


1. 'scrape_region_constraints: incoming region obligations (#52992)

589 occurrences from what looks to be the known issue #52992. Thanks to @klemens the issue contains a minimized repro. For convenience, here's also the playground link to that repro.

List of crate failures

2. unresolved inference variable in outlives: _#3t (#53568)

2 occurrences — but the 2nd crate fails because it depends on the 1st (hyper-stub) — and what looks like a possible more obscure case of #52057.

This crate is small but pretty involved with its wrappers over hyper, tokio, and futures, resulting in a lot of associated types and bounds, and some of those 'static bounds might have a weird interaction with the NLL typeck, impl Trait, and maybe delay_span_bug.

I've minimized it to only one such error, but note that the real code has 5 or 6 of them, so it would be good to verify any potential fix on the hyper-stub crate itself rather than just this minimized repro. Playground.

List of crate failures

(Marking as deferred as this is somewhat of a meta issue)

cc #53171


edit: added some missing github-only crates, going from 568 to 591 occurrences.

Metadata

Metadata

Assignees

No one assigned

    Labels

    A-NLLArea: Non-lexical lifetimes (NLL)NLL-completeWorking towards the "valid code works" goal

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions