Suggest removing value from break when invalid

[estebank]

When attempting to use break with a value in a type of loop where it'd be invalid (any non-loop), suggest using break on its own.

Close #34359.

[rust-highfive]

r? @cramertj

(rust_highfive has picked a reviewer for you, use r? to override)

[cramertj]

This second "by example" version of the error doesn't seem like it's providing the user with additional information. Would it be possible to change this PR to just add the additional help message, without the break-only example?

[estebank]

I find that when suggesting removal of text, the inline suggestion style lends itself to confusion more easily. What do you think?

[cramertj]

I find this particular line is confusing-- I don't think it adds new information, the ERROR pattern on it is misleading since there would be no error in this case (since it's the corrected code):

22 |         break //~ ERROR `break` with value from a `for` loop

[estebank]

But that's just a comment, as far as rustc is concerned. If I change the test file to have that comment one line below, it would look like this:

error[E0571]: `break` with value from a `for` loop
  --> $DIR/loop-break-value-no-repeat.rs:22:9
   |
22 |         break 22
   |         ^^^^^^^^ can only break with a value inside `loop`
help: instead, use `break` on its own without a value inside this `for` loop
   |
22 |         break
   |         ^^^^^

instead of

error[E0571]: `break` with value from a `for` loop
  --> $DIR/loop-break-value-no-repeat.rs:22:9
   |
22 |         break 22
   |         ^^^^^^^^
   |         |
   |         can only break with a value inside `loop`
   |         help: use `break` without a value: `break`

[cramertj]

Ah, okay. I personally still prefer something like this:

error[E0571]: `break` with value from a `for` loop
  --> $DIR/loop-break-value-no-repeat.rs:22:9
   |
22 |         break 22
   |         ^^^^^^^^
   |         |
   |         can only break with a value inside `loop`
   |         help: use `break` without a value inside this `for` loop

But I'll defer to your judgement.

[cramertj]

r=me with nit fixed.

[estebank]

@bors r=cramertj rollup

We can change this to inline afterwards. I'm thinking that the current output style could be reserved for -Zteach and use your preferred inline output for normal output, but until we expand/stabilize/publicize -Zteach I prefer being as explicit as possible to minimize confusion for newcomers.

[bors]

📌 Commit b7437c5 has been approved by cramertj