-
Notifications
You must be signed in to change notification settings - Fork 12.7k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
fix "bivariant wf" bug in the NLL subtyping code #54105
Comments
…n-ascription, r=pnkfelix support ascription for patterns in NLL This implements the strategy outlined in [this comment](#47184 (comment)): - We first extend the NLL subtyping code so it can handle inference variables and subtyping. - Then we extend HAIR patterns with type ascription. - Then we treat the type `T` in `let pat: T = ...` as an ascription. Before landing, a few things: - [x] Fix the WF rule bug (filed a FIXME #54105) - [x] Fix an ICE I encountered locally around bound regions, or else file a follow-up - [x] More tests probably =) r? @pnkfelix
assigning self to this to try to help offload work from @nikomatsakis |
I have yet to come up with a way to exercise this bug. I tried e.g. this example but we do catch it. |
I'm demoting this to Release. I still think there is probably a bug here but since we don't even know how to reproduce it, it doesn't seem like an RC2 blocker. |
Tagging as A-traits too because @scalexm is expanding this code as part of their work on the trait system unification code, and we may wind up fixing it then. Still need to come up with a test case. Removing from the Rust 2018 milestone. |
NLL triage. This needs prioritization. I'm inclined to go with P-medium. |
nominating for discussion at NLL meeting tonight. Namely, I'm seeking a volunteer to look for a test case. |
removing I-nominated tag. We discussed briefly at the NLL meeting but I think I didn't make the bait here attractive enough. |
NLL-triage. Marking P-medium. |
#53873 generalized the NLL subtyping code to handle subtyping and the like, but I think it is vulnerable to a similar flaw as the one from #41677. I didn't want to block that PR so I am filing this issue to follow-up.
The text was updated successfully, but these errors were encountered: