issue triage: closures, lifetimes, and nll

[memoryruins]

Most of the issues listed below appear ready to be closed or tagged. The check boxes are there to mark after a decision has been made on their respective threads (including if an issue is decided to be left open).

@pnkfelix @matthewjasper mind checking these out? what started as a general triage, ended up accumulating a bit of fixed-by-nll issues. although not all nll-related, the other issues are generally in areas you could help make a final decision on.

now compiles on both editions

rustc: 1.32.0

The following issues previously did not compile.
These appear ready to close, unless if any need a test.

• failure to infer move mode for a variable when the move is from *x failure to infer move mode for a variable when the move is from *x #23675
• Incorrect closure type deduced despite move keyword Incorrect closure type deduced despite move keyword #44336
• Rustc confuses lifetimes of different unrelated variables of type &'static str when implementing a trait for &'a str Rustc confuses lifetimes of different unrelated variables of type &'static str when implementing a trait for &'a str #26448
• Unexpected lifetime issue when comparing Option<&i32 Unexpected lifetime issue when comparing Option<&i32> #42966
• Incorrect lifetime error with a generic FnMut after annotating the type in a closure. Incorrect lifetime error with a generic FnMut after annotating the type in a closure. #22557

now compiles on 2018

rustc: 1.32.0

cc @rust-lang/wg-compiler-nll

The following issues previously did not compile, but now do on 2018.
They also compile on 2015 with#![feature(nll)].

• Weird lifetime error with closure and early-bound lifetime Weird lifetime error with closure and early-bound lifetime #28092
  • (example in description compiles in both editions, but a variant was identified that still exhibits the original problem.)
• Closures cause unnecessary borrows Closures cause unnecessary borrows #23591
• Lifetime elision is too greedy without explicit type declaration Lifetime elision is too greedy without explicit type declaration #39943
• Ref and &T have different lifetime requirements Ref<T> and &T have different lifetime requirements #37785
• if let keeps borrows too long if let keeps borrows too long #43407
• Compiler gets lifetime scope wrong Compiler gets lifetime scope wrong #28970
• Borrow checker gets confused with a mutable pointer which should be moved Borrow checker gets confused with a mutable pointer which should be moved #27487
• Slice pattern incorrectly report double borrow Slice pattern incorrectly report double borrow #37302
• Borrow checker false dropped while still borrowed Borrow checker false dropped while still borrowed #38915
• Spurious "borrowed value does not live long enough" when calling a function through a function pointer Spurious "borrowed value does not live long enough" when calling a function through a function pointer #49073

now compiles on beta / nightly on both editions

rust version 1.34.0-nightly (da6ab956e 2019-01-27)

needs test?

• Adding Deref impl affects lifetime bounds of other methods Adding Deref impl affects lifetime bounds of other methods #27697

now (correctly) rejected on beta / nightly on both editions

These appear ready to close, unless if any need a test.

• HRTB bug in streaming iterator HRTB bug in streaming iterator #30786

[memoryruins]

I've begun to comment on the individual issues with the updates and errors, as well as tagging fixed-by-nll where it seems fit (edit: I mistakenly / prematurely tagged a few as fixed-by-nll in the list, so I removed a few tags; there might be a few others to remove). Removed #30127 and #48677 from the list because they still error on 2018 when used.

[pnkfelix]

@memoryruins I believe our policy is indeed to leave issues tagged with NLL-fixed-by-NLL open, since the old borrow-checker is still on by default in the 2015 edition.

When we turn NLL on for the 2015 edition (#57804) then we will be able to start closing the NLL-fixed-by-NLL issues.

[pnkfelix]

tagging as metabug.

In general most of these issues are tagged NLL-fixed-by-NLL and so they'll tend to stay open until #57804 is addressed. Which means I don't see much point in trying to assign high priority to this issue (since most of the work items cannot be addressed until #57804 is addressed.)

[pnkfelix]

Hey lets start closing some of these that are fixed now that NLL is on for all editions in stable!

[pnkfelix]

oh wait: we cannot close the ones that correspond to errors-downgraded-to-warnings I think, because most of those are soundness issues, and we probably need to wait until we stop the downgrading-to-warnings before we close all those issues. 😢

[crlf0710]

@pnkfelix #30786 has test added in #62519 Maybe we can tick the last box... and maybe close this issue?