nll: respect user type annotations with constants in patterns

[nikomatsakis]

Breaking off from #47184 -- we need to respect user type annotations in patterns. Here are some examples:

Compiles without NLL (playground):

trait Foo<'a> {
    const C: &'a u32;
}

impl<'a, T> Foo<'a> for T {
    const C: &'a u32 = &22;
}

fn foo() {
    let a = 22;
    match &a {
        <() as Foo<'static>>::C => { }
        &_ => { }
    }
}

fn main() {
}

Errors without NLL (playground):

use std::cell::Cell;

trait Foo<'a> {
    const C: Option<Cell<&'a u32>>;
}

impl<'a, T> Foo<'a> for T {
    const C: Option<Cell<&'a u32>> = None;
}

fn foo() {
    let a = 22;
    let b = Some(Cell::new(&a));
    match b {
        <() as Foo<'static>>::C => { }
        _ => { }
    }
}

fn main() {
}

[pnkfelix]

triage: Either this is something that we're planning to implement and backport to the beta channel... or its something that is not that high priority.

Given that this is an NLL-sound issue, I'm going to assume that it is indeed high priority and belongs on the Release milestone.

Thus, tagging as P-high

[pnkfelix]

Just a quick note: this variant of the description's second example is rejected by NLL (play):

#![feature(nll)]

use std::cell::Cell;

trait Foo<'a> {
    const C: Option<Cell<&'a u32>>;
}

impl<'a, T> Foo<'a> for T {
    const C: Option<Cell<&'a u32>> = None;
}

fn foo<'a>(r: &'a i32) {
    let b = Some(Cell::new(r));
    match b {
        <() as Foo<'static>>::C => { }
        _ => { }
    }
}

fn main() {
}

From looking at the MIR generated for the description's second example, perhaps the problem stems from this ?

    let mut _4: &'static u32;
...
        StorageLive(_1);                 // bb0[0]: scope 0 at ../../should-cfail.rs:12:9: 12:10
        _1 = const 22u32;                // bb0[1]: scope 0 at ../../should-cfail.rs:12:13: 12:15
...
        _4 = &'static _1;                // bb0[6]: scope 1 at ../../should-cfail.rs:13:28: 13:30

(I got the 'static annotation in the _4 = &'static _1; via -Z identify-regions. Though I'm not 100% sure that debug flag reflects the NLL state. Anyway, it might be important to make that case work.)

[pnkfelix]

triage: assigning to self

[pnkfelix]

didn't get a chance to look into this further today. Unassigning self (I'll be back from leave in ~2 weeks, but I want to make sure other people know that this is available to hack on in the meantime).

[nikomatsakis]

Removing from milestone — sufficiently obscure that it need not be a release blocker