NLL diagnostics fail to report lifetime mismatches

[pnkfelix]

The following tests report the lifetime mismatches nicely under AST borrowck but fail to do so under NLL.

Tests:

• lifetime-errors/ex3-both-anon-regions-2.rs

(This list of tests is drawn from an informal paper document that I have been using to keep notes for myself as I work on this...)

Example (ex3-both-anon-regions-2.rs)

AST diagnostic:

error[E0623]: lifetime mismatch
  --> $DIR/ex3-both-anon-regions-2.rs:12:9
   |
LL | fn foo((v, w): (&u8, &u8), x: &u8) {
   |                 ---           --- these two types are declared with different lifetimes...
LL |     v = x; //~ ERROR lifetime mismatch
   |         ^ ...but data from `x` flows here

NLL diagnostic:

warning: not reporting region error due to nll
  --> $DIR/ex3-both-anon-regions-2.rs:12:9
   |
LL |     v = x; //~ ERROR lifetime mismatch
| ^

---

Update (pnkfelix)

• Wait, is this assignment actually legal under NLL? Seems like it shouldn't be legal. But the test could be made more robust to actually illustrate why such an assignment would be broken.

[pnkfelix]

I don't think the assignment as written should be broken under NLL. its not like we're mutating the tuple in the way that the caller can observe.

If this were a &mut (ref mut v, w): &mut (&u8, &u8), x: &u8, then we might be going somewhere, right?

[pnkfelix]

I ended up adopting the aforementioned form of the test in #51057