Point at : when using it instead of ;

[estebank]

When triggering type ascription in such a way that we can infer a
statement end was intended, add a suggestion for the change. Always
point out the reason for the expectation of a type is due to type
ascription.

Fix #42057, #41928.

[rust-highfive]

r? @nikomatsakis

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

[bors]

☔ The latest upstream changes (presumably #43060) made this pull request unmergeable. Please resolve the merge conflicts.

[KalitaAlexey]

Will it work if a colon was here?

[estebank]

The output in that case is

error: expected type, found `0`
  --> ../../src/test/ui/suggestions/type-ascription-instead-of-statement-end.rs:15:5
   |
14 |     println!("test"):
   |                     - help: did you mean to use `;` here instead?
15 |     0:
   |     ^ expecting a type here because of type ascription

[nikomatsakis]

this sort of looks funny, perhaps because of the ^ having no label. I think I'd prefer:

error: expected type, found `0`
  --> $DIR/type-ascription-instead-of-statement-end.rs:15:5
   |
14 |     println!("test"):
   |                     - help: did you mean to use a `;` here?
15 |     0;
   |     ^ expecting a type here because of type ascription

error: aborting due to previous error

[estebank]

Done.

[nikomatsakis]

Huh, this looks strange to me somehow. Why not say "help: did you mean to use ; here?"? I think users might not know that "end the statement" means ; -- but also, I am not sure about this "floating" ; that appears at the end of the sentence.

[estebank]

The presentation is because it is an inline suggestion. Should I change the code so that we can control wether the code to be replaced should be displayed when inline?

[nikomatsakis]

Hmm; I think either change the code, or change the wording. But yeah probably just having the option to "suppress" the suggestion when displayed inline seems good.

[estebank]

Done.

[shepmaster]

This is your polite 7-day ping @estebank — when do you think you'll be able to respond to the comments?

[nikomatsakis]

Failure:

[01:08:34] failures:
[01:08:34]     [ui] ui/issue-22644.rs

The new message looks improved, though. =)

[estebank]

@nikomatsakis fixed

[estebank]

@nikomatsakis covered another case that fails during resolve instead of parse:

fn f() {}

fn main() {
    f():
    f();
}

[estebank]

@nikomatsakis ping

[nikomatsakis]

@bors r+

[bors]

📌 Commit e39bcec has been approved by nikomatsakis

[nikomatsakis]

cc @oli-obk, I don't this suggestion follows the guidelines, but I also think it looks good. Perhaps we need to adjust the guidelines to account for the "suppress example code" option?

[bors]

⌛ Testing commit e39bcec with merge afe145d...

[oli-obk]

I'm assuming you mean that the guildelines aren't followed by the use of did you mean. This guideline exists because of #42033 (comment) which notes that the did you mean sounds condescending. I don't have a strong opinion on this guideline though.

"Try using a ; instead:" looks fine to me. Otherwise let's adjust the guidelines if that's desired.

[bors]

☀️ Test successful - status-appveyor, status-travis
Approved by: nikomatsakis
Pushing afe145d to master...

[luser]

@estebank Thanks for making this error message better!

[Nokel81]

Which error message got finalized?

[estebank]

@Nokel81 sorry, I don't understand the question. Could you rephrase? This PR adds a label stating "help: did you mean to use ; here?" when type ascription (triggered by :) ends the previous line to the type error, as it probably was a mistyped :

[Nokel81]

I was talking about the guideline of not using did you mean that was referenced several times near the end of this PR chain. Was the guidelines followed or was the original error message kept as help: did you mean to use ; here instead?

[estebank]

@Nokel81 ah, I see. No, I didn't change the message before it got merged. I can post a PR later today/tomorrow changing the wording, if you want.

[Nokel81]

No it is fine, I personally feel that there are some cases where did you mean makes sense. Mostly for small syntactic typos like this. I was just wondering if it had actually been changed