Add information to higher-ranked lifetimes conflicts error messages #57901

lqd · 2019-01-25T18:45:57Z

Make these errors go through the new "placeholder error" code path, to have self tys displayed and make them hopefully less confusing.

Should fix #57362.

r? @nikomatsakis — so we can iterate on the specific wording you wanted.

Centril · 2019-01-25T19:08:30Z

src/test/ui/issues/issue-57362.rs

+    fn f(self);
+}
+
+impl<T> Trait for fn(&T) {


I'd write this out:

Suggested change

impl<T> Trait for fn(&T) {

impl<T> Trait for for<'a> fn(&'a T) {

to make what's tested clearer.

src/test/ui/issues/issue-57362.stderr

Centril · 2019-01-25T19:18:33Z

src/test/ui/issues/issue-57362.stderr

+LL |     a.f(); //~ ERROR not general enough
+   |       ^
+   |
+   = note: `Trait` would have to be implemented for the type `fn(&u8)`


Suggested change

= note: `Trait` would have to be implemented for the type `fn(&u8)`

= note: `Trait` must be implemented for the type `fn(&u8)`

Shorter and more direct :)
(alternatively s/must/needs to/)

I'm not so sure. I find the word "must" more confusing somehow. "would have to be" makes it clearer to me that it is not the case.

In that case: "Trait isn't, but must be..."; or "Trait needed to be..." -- I just find the current text wordy, but its not a big deal or anything. :)

src/librustc/infer/error_reporting/nice_region_error/placeholder_error.rs

Centril · 2019-01-25T19:26:49Z

src/librustc/infer/error_reporting/nice_region_error/placeholder_error.rs

+                    if self_ty_has_vid {
+                        err.note(&format!(
+                            "but `{}` is actually implemented for the type `{}`, \
+                            for the specific lifetime `'{}`",


Nothing to do about here... but it would have been nice to "switch" on the format in some way since the only difference is the format and not the arguments... I suspect this contributes to duplication in many places...

I guess you mean moving "the specific" and "some" into some string? I'm of mixed minds. I often prefer to read the full text. It's also more friendly to i18n if we ever do that (ha, that's a long way away). But OTOH, it is duplication indeed. No strong opinion.

I guess you mean moving "the specific" and "some" into some string?

Right; I was mainly thinking of switching on the "string literal"; but that doesn't work with format! afaik so its probably not implementable? (which is why it mostly was a general comment about the future... ^^)

src/test/ui/hrtb/hrtb-cache-issue-54302.stderr

lqd · 2019-01-26T20:45:56Z

Updated to address the review comments here and on Zulip:

The code handling all the notes has been extracted to a helper function
All the cases needed to correctly print either the trait or the self type where the sub/sup placeholders appear next to these placeholders, and not at opposite ends of the error message, have been added to this helper function.

Some errors (e.g placeholder errors) have unresolved type vars so this will allow to use `resolve_type_vars_if_possible` when needed.

…e `sup` region These are happening since the switch to universes, and will now go through the "placeholder error" path, instead of the current fallback of E308 "mismatched types" errors.

These can sometimes be unresolved: some of the rustc UI tests show this.

…closer to the lifetimes When mentioning lifetimes, only invert wording between the expected trait and the self type when the self type has the vid. This way, the lifetimes always stay close to the self type or trait ref that actually contains them.

…r fn

src/test/ui/issues/issue-57362-2.stderr

…ther than "the specific lifetime"

nikomatsakis · 2019-01-29T14:22:29Z

@bors r+

bors · 2019-01-29T14:22:30Z

📌 Commit c97d135 has been approved by nikomatsakis

nikomatsakis · 2019-01-29T14:22:45Z

@bors p=1 -- fairly severe, if limited, diagnostic regression

bors · 2019-01-29T16:58:23Z

⌛ Testing commit c97d135 with merge 7425663...

@nikomatsakis

Add information to higher-ranked lifetimes conflicts error messages Make these errors go through the new "placeholder error" code path, to have self tys displayed and make them hopefully less confusing. Should fix #57362. r? @nikomatsakis — so we can iterate on the specific wording you wanted.

bors · 2019-01-29T19:40:41Z

☀️ Test successful - checks-travis, status-appveyor
Approved by: nikomatsakis
Pushing 7425663 to master...

bors · 2019-01-29T19:42:16Z

☀️ Test successful - checks-travis, status-appveyor
Approved by: nikomatsakis
Pushing 7425663 to master...

pnkfelix · 2019-01-30T10:50:41Z

beta nominating so that we discuss at T-compiler meeting. (But to be clear, I personally am skeptical about trying to backport this. We can probably live with a diagnostic regression, even one this severe, in the stable channel for a release cycle.)

pnkfelix · 2019-01-31T15:36:42Z

discussed at T-compiler meeting. declining for beta backport.

rust-highfive assigned nikomatsakis Jan 25, 2019

rust-highfive added the S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. label Jan 25, 2019

This comment has been minimized.

Sign in to view

Centril reviewed Jan 25, 2019

View reviewed changes

nikomatsakis reviewed Jan 25, 2019

View reviewed changes

src/test/ui/hrtb/hrtb-cache-issue-54302.stderr Outdated Show resolved Hide resolved

This comment has been minimized.

Sign in to view

lqd and others added 15 commits January 27, 2019 10:52

Make NiceRegionError use the InferCtxt instead of its TyCtxt

c1437c9

Some errors (e.g placeholder errors) have unresolved type vars so this will allow to use `resolve_type_vars_if_possible` when needed.

Handle higher-ranked lifetime conflict errors where the subtype is th…

a602826

…e `sup` region These are happening since the switch to universes, and will now go through the "placeholder error" path, instead of the current fallback of E308 "mismatched types" errors.

Try to resolve type vars in the placeholder errors trait refs

823c888

These can sometimes be unresolved: some of the rustc UI tests show this.

Try to reword placeholder error messages to make them clearer

55389f9

Update two E308 tests to the new placeholder error

ce61b1b

Update test expectations for new placeholder error messages

a79f135

Test new placeholder error messages in previously untested combinations

f5a74d4

break apart tests

c5dea57

add a lot more debug! to try_report_placeholders_trait

ab80162

identify when implemented for "some specific lifetime"

ec6405b

comment the pattern

790ed91

Fix issue-57362-1.rs attributes

1730ad4

Extract the notes explaining the placeholder region errors to a helpe…

43c0518

…r fn

Fix stray typo

e077501

lqd force-pushed the issue_57362 branch from d01879e to e077501 Compare January 27, 2019 09:54

nikomatsakis reviewed Jan 28, 2019

View reviewed changes

src/test/ui/issues/issue-57362-2.stderr Outdated Show resolved Hide resolved

Refer to synthetically named lifetimes as "some specific lifetime" ra…

c97d135

…ther than "the specific lifetime"

nikomatsakis approved these changes Jan 29, 2019

View reviewed changes

bors removed the S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. label Jan 29, 2019

bors added the S-waiting-on-bors Status: Waiting on bors to run and complete tests. Bors will change the label on completion. label Jan 29, 2019

bors merged commit c97d135 into rust-lang:master Jan 29, 2019

bors mentioned this pull request Jan 29, 2019

Introduce Rust symbol mangling scheme. #57967

Merged

lqd deleted the issue_57362 branch January 29, 2019 20:03

pnkfelix added beta-nominated Nominated for backporting to the compiler in the beta channel. T-compiler Relevant to the compiler team, which will review and decide on the PR/issue. labels Jan 30, 2019

pnkfelix mentioned this pull request Jan 30, 2019

[NLL] Bad higher ranked subtype error #57374

Open

pnkfelix removed the beta-nominated Nominated for backporting to the compiler in the beta channel. label Jan 31, 2019

pnkfelix mentioned this pull request Feb 15, 2019

make generalization code create new variables in correct universe #58056

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Add information to higher-ranked lifetimes conflicts error messages #57901

Add information to higher-ranked lifetimes conflicts error messages #57901

lqd commented Jan 25, 2019

This comment has been minimized.

Centril Jan 25, 2019

Centril Jan 25, 2019

nikomatsakis Jan 25, 2019

Centril Jan 25, 2019

Centril Jan 25, 2019

nikomatsakis Jan 25, 2019

Centril Jan 25, 2019

This comment has been minimized.

lqd commented Jan 26, 2019

This comment has been minimized.

nikomatsakis commented Jan 29, 2019

bors commented Jan 29, 2019

nikomatsakis commented Jan 29, 2019

bors commented Jan 29, 2019

bors commented Jan 29, 2019

bors commented Jan 29, 2019

pnkfelix commented Jan 30, 2019 •

edited

Loading

pnkfelix commented Jan 31, 2019

	impl<T> Trait for fn(&T) {
	impl<T> Trait for for<'a> fn(&'a T) {

	= note: `Trait` would have to be implemented for the type `fn(&u8)`
	= note: `Trait` must be implemented for the type `fn(&u8)`

Add information to higher-ranked lifetimes conflicts error messages #57901

Add information to higher-ranked lifetimes conflicts error messages #57901

Conversation

lqd commented Jan 25, 2019

This comment has been minimized.

Centril Jan 25, 2019

Choose a reason for hiding this comment

Centril Jan 25, 2019

Choose a reason for hiding this comment

nikomatsakis Jan 25, 2019

Choose a reason for hiding this comment

Centril Jan 25, 2019

Choose a reason for hiding this comment

Centril Jan 25, 2019

Choose a reason for hiding this comment

nikomatsakis Jan 25, 2019

Choose a reason for hiding this comment

Centril Jan 25, 2019

Choose a reason for hiding this comment

This comment has been minimized.

lqd commented Jan 26, 2019

This comment has been minimized.

nikomatsakis commented Jan 29, 2019

bors commented Jan 29, 2019

nikomatsakis commented Jan 29, 2019

bors commented Jan 29, 2019

bors commented Jan 29, 2019

bors commented Jan 29, 2019

pnkfelix commented Jan 30, 2019 • edited Loading

pnkfelix commented Jan 31, 2019

pnkfelix commented Jan 30, 2019 •

edited

Loading