Handle generic bounds in a uniform way in HIR #93803

cjgillot · 2022-02-09T12:16:32Z

Generic bounds in HIR used to be split between bounds in the parameter definition and bounds in a where clause. This PR attempts to store all of those as where predicates.

This effectively desugars

fn foo<T: Default, U>(x: impl Copy) where U: Clone

into

fn foo<T, U, _V>(x: _V) where T: Default, U: Clone, _V: Copy

(where _V is actually hidden and called "impl Copy").

I managed to make compiler warnings more uniform.
About rustdoc: is making this desugaring user-visible acceptable?
About clippy: I don't understand the subtle logic in the needless-lifetimes lint.

r? @estebank

rust-highfive · 2022-02-09T12:16:35Z

Some changes occurred in clean/types.rs.

cc @camelid

rustdoc-json-types is a public (although nightly-only) API. Consider changing src/librustdoc/json/conversions.rs instead; otherwise, make sure you update format_version.

cc @CraftSpider,@aDotInTheVoid

Some changes occurred in src/tools/clippy.

cc @rust-lang/clippy

petrochenkov · 2022-02-09T12:21:02Z

About rustdoc: is making this desugaring user-visible acceptable?

I'm not sure how important this is, but AFAIR this is the primary reason why it wasn't done before.

CraftSpider · 2022-02-09T16:02:20Z

I'm not sure I'd mind seeing all bounds as where, but I'd definitely defer to other rustdoc members on that. If this is going to merge though, please ensure you also bump FORMAT_VERSION in the rustdoc-json-types module. It needs to be increased on every breaking change.

camelid · 2022-02-09T20:02:20Z

I'm not sure I'd mind seeing all bounds as where, but I'd definitely defer to other rustdoc members on that

I've kind of thought of doing this anyway, so I'm slightly in favor. This would also improve consistency since for some reason, cross-crate re-exports always display their bounds as where clauses IIRC. Maybe because the ty representation already has this desugaring?

aDotInTheVoid · 2022-02-09T21:23:19Z

I think its important to show that their is a difference in rustdoc html, as impl Trait cannot be used with the turbofish, although AFAIKT the pr still does that.

The JSON output also has a significant regression in that pub fn by_impl_trait(x: impl Foo) {} is given arguments ["x", {"inner": "impl Foo", "kind": "generic"}], and no mension of the id of Foo anywhere.

Unrelated, but we should probably merge src/etc/check_missing_items.py into jsondocck

camelid · 2022-02-10T20:29:41Z

impl Trait should definitely not be shown as a generic.

cjgillot · 2022-02-10T21:01:39Z

@aDotInTheVoid how do I run the json tests?

src/test/ui/associated-type-bounds/type-alias.stderr

src/test/ui/generic-associated-types/issue-86483.stderr

src/test/ui/impl-trait/must_outlive_least_region_or_bound.stderr

Urgau · 2022-02-10T23:52:20Z

@aDotInTheVoid how do I run the json tests?

./x.py test src/test/rustdoc-json

src/test/ui/suggestions/suggest-impl-trait-lifetime.stderr

estebank

I need to go over the code changes, but the output changes look reasonable to me (with some nitpicking for things we could improve) so it's likely that the code changes themselves are too. I'll read the changes now.

estebank · 2022-02-11T00:58:54Z

I agree with @camelid, if the signature uses impl Trait, then the docs and error messages need to refer to impl Trait and not X: Trait in all cases. The internal representation can change so that they are handled more gracefully though.

flip1995 · 2022-02-11T08:40:06Z

About clippy: I don't understand the subtle logic in the needless-lifetimes lint.

I don't see any changes to Clippy's tests. The changes to lifetimes.rs seem reasonable AFAICT. Do you get test failures in Clippy?

cjgillot · 2022-02-11T17:05:53Z

I agree with @camelid, if the signature uses impl Trait, then the docs and error messages need to refer to impl Trait and not X: Trait in all cases. The internal representation can change so that they are handled more gracefully though.

Of course, the desugaring will not appear in docs and errors. Only the internal representation is changed by this PR.

cjgillot · 2022-02-11T17:06:58Z

src/tools/clippy/tests/ui/extra_unused_lifetimes.stderr

@@ -6,12 +6,6 @@ LL | fn unused_lt<'a>(x: u8) {}
   |
   = note: `-D clippy::extra-unused-lifetimes` implied by `-D warnings`

-error: this lifetime isn't used in the function definition


@flip1995 I mean this change, and the one below. I don't understand how clippy use to account for lifetime uses in bounds and not where clauses.

I can't really tell what would cause this from just looking at the Clippy changes. I'll have to review and understand the other changes to rustc in this PR. This might take me a few days until I get to that though.

bors · 2022-02-11T20:44:09Z

☔ The latest upstream changes (presumably #93893) made this pull request unmergeable. Please resolve the merge conflicts.

rust-timer · 2022-04-30T17:50:59Z

Finished benchmarking commit (d201c81): comparison url.

Summary:

Primary benchmarks: 😿 relevant regressions found
Secondary benchmarks: 😿 relevant regressions found

	Regressions 😿 (primary)	Regressions 😿 (secondary)	Improvements 🎉 (primary)	Improvements 🎉 (secondary)	All 😿 🎉 (primary)
count¹	11	4	0	0	11
mean²	0.3%	0.3%	N/A	N/A	0.3%
max	0.5%	0.3%	N/A	N/A	0.5%

If you disagree with this performance assessment, please file an issue in rust-lang/rustc-perf.

Next Steps: If you can justify the regressions found in this perf run, please indicate this with @rustbot label: +perf-regression-triaged along with sufficient written justification. If you cannot justify the regressions please open an issue or create a new PR that fixes the regressions, add a comment linking to the newly created issue or PR, and then add the perf-regression-triaged label to this PR.

@rustbot label: +perf-regression

number of relevant changes ↩
the arithmetic mean of the percent change ↩

Mark-Simulacrum · 2022-05-03T12:31:44Z

@cjgillot It looks like the pre-merge perf justification in #93803 (comment) was that the improvements balanced regressions, but post-merge we are not seeing any improvements. I think the (presumably) improve code quality/maintenance may still justify this change, but it would be good to re-assess from your perspective.

jackh726 · 2022-05-06T22:01:32Z

I must have missed this or forgotten about it.

This seems counterintuitive to me in the context of the TAIT refactoring. Of course, @oli-obk knows more than me on this, but surprising nonetheless.

…ession, r=notriddle Fix jump to def regression rust-lang#93803 introduced a regression in the "jump to def" feature. This fixes it. Nice side-effect: it adds a new regression test. :) I also used this opportunity to add documentation about this unstable feature in the rustdoc book. cc `@cjgillot` r? `@notriddle`

…ession, r=notriddle Fix jump to def regression rust-lang#93803 introduced a regression in the "jump to def" feature. This fixes it. Nice side-effect: it adds a new regression test. :) I also used this opportunity to add documentation about this unstable feature in the rustdoc book. cc ``@cjgillot`` r? ``@notriddle``

With rust-lang#93803 `impl Trait` function arguments get desugared to hidden where bounds. However, Clippy needs to know if a bound was originally a impl Trait or an actual bound. This adds a field to the `WhereBoundPredicate` struct to keep track of this information during HIR lowering.

…cjgillot Track if a where bound comes from a impl Trait desugar With rust-lang#93803 `impl Trait` function arguments get desugared to hidden where bounds. However, Clippy needs to know if a bound was originally a `impl Trait` or an actual bound. This adds a field to the `WhereBoundPredicate` struct to keep track of this information during AST->HIR lowering. r? `@cjgillot` cc `@estebank` (as the reviewer of rust-lang#93803)

…-errors Fortify handing of where bounds on trait & trait alias definitions Closes rust-lang#96664 Closes rust-lang#96665 Since rust-lang#93803, when listing all bounds and predicates we now need to account for the possible presence of predicates on any of the generic parameters. Both bugs were hidden by the special handling of bounds at the generic parameter declaration position. Trait alias expansion used to confuse predicates on `Self` and where predicates. Exiting too late when listing all the bounds caused a cycle error.

Track if a where bound comes from a impl Trait desugar With rust-lang/rust#93803 `impl Trait` function arguments get desugared to hidden where bounds. However, Clippy needs to know if a bound was originally a `impl Trait` or an actual bound. This adds a field to the `WhereBoundPredicate` struct to keep track of this information during AST->HIR lowering. r? `@cjgillot` cc `@estebank` (as the reviewer of #93803)

With rust-lang#93803 `impl Trait` function arguments get desugared to hidden where bounds. However, Clippy needs to know if a bound was originally a impl Trait or an actual bound. This adds a field to the `WhereBoundPredicate` struct to keep track of this information during HIR lowering.

rust-highfive assigned estebank Feb 9, 2022

rustbot added T-compiler Relevant to the compiler team, which will review and decide on the PR/issue. T-rustdoc Relevant to the rustdoc team, which will review and decide on the PR/issue. labels Feb 9, 2022

rust-highfive added the S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. label Feb 9, 2022

This comment has been minimized.

Sign in to view

This comment was marked as outdated.

Sign in to view

cjgillot force-pushed the unify-bounds branch from ff6cc12 to 3bb9c89 Compare February 10, 2022 21:00

This comment has been minimized.

Sign in to view

estebank reviewed Feb 10, 2022

View reviewed changes

src/test/ui/associated-type-bounds/type-alias.stderr Outdated Show resolved Hide resolved

estebank reviewed Feb 10, 2022

View reviewed changes

src/test/ui/generic-associated-types/issue-86483.stderr Outdated Show resolved Hide resolved

estebank reviewed Feb 10, 2022

View reviewed changes

src/test/ui/impl-trait/must_outlive_least_region_or_bound.stderr Outdated Show resolved Hide resolved

estebank reviewed Feb 11, 2022

View reviewed changes

src/test/ui/suggestions/suggest-impl-trait-lifetime.stderr Outdated Show resolved Hide resolved

estebank reviewed Feb 11, 2022

View reviewed changes

cjgillot commented Feb 11, 2022

View reviewed changes

This comment has been minimized.

Sign in to view

cjgillot force-pushed the unify-bounds branch from f9089ca to ff050e4 Compare February 11, 2022 22:56

This comment was marked as outdated.

Sign in to view

cjgillot force-pushed the unify-bounds branch from ff050e4 to c3a9feb Compare February 14, 2022 19:36

aDotInTheVoid mentioned this pull request Apr 30, 2022

New Release aDotInTheVoid/rustdoc-types#8

Closed

GuillaumeGomez added a commit to GuillaumeGomez/rust that referenced this pull request May 2, 2022

Fix regression in link-to-definition introduced in rust-lang#93803

a26cb61

GuillaumeGomez mentioned this pull request May 2, 2022

Fix jump to def regression #96636

Merged

Edgeworth mentioned this pull request May 3, 2022

Regression in nightly-2022-05-01 relating to trait objects #96664

Closed

Mark-Simulacrum removed the perf-regression-triaged The performance regression has been triaged. label May 3, 2022

cjgillot mentioned this pull request May 3, 2022

Render all trait bounds in where clauses in rustdoc #96680

Closed

lqd mentioned this pull request May 4, 2022

cycle detected when computing the super predicates of ... #96665

Closed

cjgillot mentioned this pull request May 4, 2022

Fortify handing of where bounds on trait & trait alias definitions #96715

Merged

flip1995 mentioned this pull request May 6, 2022

Track if a where bound comes from a impl Trait desugar #96770

Merged

apiraino mentioned this pull request Jun 7, 2022

"warning: unnecessary lifetime parameter" unsilenceable lint #96956

Closed

compiler-errors mentioned this pull request Jun 20, 2022

Syntactically invalid suggestion for "help: consider further restricting this bound" #93828

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Handle generic bounds in a uniform way in HIR #93803

Handle generic bounds in a uniform way in HIR #93803

cjgillot commented Feb 9, 2022 •

edited

Loading

rust-highfive commented Feb 9, 2022

petrochenkov commented Feb 9, 2022

This comment has been minimized.

CraftSpider commented Feb 9, 2022

camelid commented Feb 9, 2022 •

edited

Loading

aDotInTheVoid commented Feb 9, 2022

This comment was marked as outdated.

camelid commented Feb 10, 2022

cjgillot commented Feb 10, 2022

This comment has been minimized.

Urgau commented Feb 10, 2022

estebank left a comment

estebank commented Feb 11, 2022

flip1995 commented Feb 11, 2022

cjgillot commented Feb 11, 2022

cjgillot Feb 11, 2022

flip1995 Feb 16, 2022

This comment has been minimized.

bors commented Feb 11, 2022

This comment was marked as outdated.

rust-timer commented Apr 30, 2022

Mark-Simulacrum commented May 3, 2022

jackh726 commented May 6, 2022

Handle generic bounds in a uniform way in HIR #93803

Handle generic bounds in a uniform way in HIR #93803

Conversation

cjgillot commented Feb 9, 2022 • edited Loading

rust-highfive commented Feb 9, 2022

petrochenkov commented Feb 9, 2022

This comment has been minimized.

CraftSpider commented Feb 9, 2022

camelid commented Feb 9, 2022 • edited Loading

aDotInTheVoid commented Feb 9, 2022

This comment was marked as outdated.

camelid commented Feb 10, 2022

cjgillot commented Feb 10, 2022

This comment has been minimized.

Urgau commented Feb 10, 2022

estebank left a comment

Choose a reason for hiding this comment

estebank commented Feb 11, 2022

flip1995 commented Feb 11, 2022

cjgillot commented Feb 11, 2022

cjgillot Feb 11, 2022

Choose a reason for hiding this comment

flip1995 Feb 16, 2022

Choose a reason for hiding this comment

This comment has been minimized.

bors commented Feb 11, 2022

This comment was marked as outdated.

rust-timer commented Apr 30, 2022

Footnotes

Mark-Simulacrum commented May 3, 2022

jackh726 commented May 6, 2022

cjgillot commented Feb 9, 2022 •

edited

Loading

camelid commented Feb 9, 2022 •

edited

Loading