Duplicate bounds lint

[ibraheemdev]

Deprecate clippy::trait_duplication_in_bounds and add a rustc built-in lint duplicate_bounds that handles more cases:

• fn foo<T: Foo + Foo>() {}
• fn foo<T: Foo>() where T: Foo {}
• fn foo<'a, 'b: 'a + 'a>() {}
• fn foo<'a, 'b: 'a>() where 'b: 'a {}
• fn foo<'a, T: 'a>() where T: 'a {}
• (and the same for structs, impl blocks, etc.)

Resolves #88013.

[rust-highfive]

Some changes occurred in src/tools/clippy.

cc @rust-lang/clippy

[rust-highfive]

r? @petrochenkov

(rust-highfive has picked a reviewer for you, use r? to override)

[ibraheemdev]

This seems to be a false positive from the old clippy lint as well:

#![deny(clippy::pedantic)]

fn foo<T: AsRef<str>>()
where
    T: AsRef<String>,
{
}

error: this trait bound is already specified in the where clause
 --> src/lib.rs:3:11
  |
3 | fn foo<T: AsRef<str>>()
  |           ^^^^^^^^^^
  |

I guess I would have to compare Path::segments instead of Res to handle this case?

[camsteffen]

Can you add a Clippy test for the renamed lint in tests/ui/rename.rs?

[ibraheemdev]

In addition to the test in src/tools/clippy/tests/ui/deprecated.rs?

[camsteffen]

In addition to the test in src/tools/clippy/tests/ui/deprecated.rs?

Oh, no that's fine, since other renames are also tested there. But there should be more output in deprecated.stderr.

[rust-log-analyzer]

The job x86_64-gnu-llvm-10 failed! Check out the build log: (web) (plain)

Click to see the possible cause of the failure (guessed by this bot)

   Compiling rustc_mir_build v0.0.0 (/checkout/compiler/rustc_mir_build)
   Compiling rustc_mir v0.0.0 (/checkout/compiler/rustc_mir)
   Compiling rustc_typeck v0.0.0 (/checkout/compiler/rustc_typeck)
   Compiling rustc_plugin_impl v0.0.0 (/checkout/compiler/rustc_plugin_impl)
error: this trait bound has already been specified
    |
    |
222 |         F: Float + Into<Scalar<M::PointerTag>> + FloatConvert<Single> + FloatConvert<Double>,
    |
    |
    = note: `-D duplicate-bounds` implied by `-D warnings`
    = help: consider removing this trait bound
error: could not compile `rustc_mir` due to previous error
warning: build failed, waiting for other jobs to finish...
error: build failed
Build completed unsuccessfully in 0:08:50

[petrochenkov]

Suggested change

	ls.register_renamed("clippy::trait_duplication_in_bounds", "trait_duplication_in_bounds");
	ls.register_renamed("clippy::trait_duplication_in_bounds", "duplicate_bounds");

[petrochenkov]

Why is the lint reported here?

[ibraheemdev]

Because Try and TryV2 are the same thing now. cc @scottmcm

[scottmcm]

Oh, good point, I never cleaned that alias up after the bootstrap compiler updated. New PR: #88223

[petrochenkov]

I'm sufficiently sure that the check should not be HIR-based.

Before type checking all bounds are converted to type representation (struct Predicate or something like that).
IIRC, they are "normalized" and interned in process (with aliases removed etc.), making semantic bound comparisons easy.
The lint should use that type checker representation and semantic comparisons, IMO.
The Trait<A> + Trait<B> false positive in particular is unacceptable.

(Could you also add a test for T: Trait + TraitAlias where trait TraitAlias = Trait; after migrating to type checker representation?)

[bors]

☔ The latest upstream changes (presumably #88329) made this pull request unmergeable. Please resolve the merge conflicts.

[JohnCSimon]

Ping from triage:
@ibraheemdev Can you please resolve the merge conflict?

[JohnCSimon]

Ping again from triage:
@ibraheemdev Can you please resolve the merge conflict?

[JohnCSimon]

Ping from triage:
@ibraheemdev - this PR hasn't seen any movement in over a month, so I'm closing this as inactive. Please feel free to reopen when you are ready to continue.