Fix ICE in label_fn_like for splatted args and add UI tests

[Ajay-singh1]

This PR builds on top of #153697 (teor's #[splat] implementation).

Bug Fix

While writing UI tests for the unhappy path of the overload-at-home example, I discovered an ICE when calling a splatted function with an argument combination that has no registered impl:

foo.method(42i32, 42i32); // no impl for (i32, i32)
// ICE: arg index 1 out of bounds for method with 1 inputs

The condition !tuple_arguments.is_splatted() in label_fn_like was inverted , the safe use_splat_fallback path ran for non-splatted functions, and the span_bug! fired for splatted ones. Fix is removing the !.

After the fix the compiler gives a clean error instead:

error[E0308]: mismatched types expected String, found i32

CC: @teor2345

[rustbot]

Some changes occurred in compiler/rustc_hir/src/attrs

cc @jdonszelmann, @JonathanBrouwer

Some changes occurred in compiler/rustc_attr_parsing

cc @jdonszelmann, @JonathanBrouwer

rust-analyzer is developed in its own repository. If possible, consider making this change to rust-lang/rust-analyzer instead.

cc @rust-lang/rust-analyzer

HIR ty lowering was modified

cc @fmease

Some changes occurred to the CTFE / Miri interpreter

cc @rust-lang/miri

The reflection data structures are tied exactly to the implementation
in the compiler. Make sure to also adjust rustc_const_eval/src/const_eval/type_info.rs

cc @oli-obk

Some changes occurred to the CTFE machinery

cc @RalfJung, @oli-obk, @lcnr

The Miri subtree was changed

cc @rust-lang/miri

This PR changes rustc_public

cc @oli-obk, @celinval, @ouz-a, @makai410

Some changes occurred in compiler/rustc_passes/src/check_attr.rs

cc @jdonszelmann, @JonathanBrouwer

The Cranelift subtree was changed

cc @bjorn3

[rustbot]

Failed to set assignee to teor2345: invalid assignee

Note: Only org members with at least the repository "read" role, users with write permissions, or people who have commented on the PR may be assigned.

[RalfJung]

I'd recommend opening such PRs as a draft until they are ready for review. That avoids pinging everyone. :)

[teor2345]

r? teor2345

[teor2345]

I cherry-picked the fix and extra tests to #153697, so any code either of us write on top of that PR is run on the full set of tests. I also made the minor comment fixes I suggested in commit 8d7a905

[teor2345]

Looks good, I have a nitpick on comments, and a suggestion for related tests.

If we can, let's try to keep failing and passing tests in different files. (I know I haven't always done that, if you want you can split existing test files into passing and failing.)

View changes since this review

[teor2345]

Nitpick: this comment is useful, but it would be better to edit that file and remove the FIXME (and this comment).

Same for the other files in this commit.

[Ajay-singh1]

Sure! I'll push a commit ASAP.

[teor2345]

Suggestion:

Can we also test splatting impl Tuple arguments?
https://doc.rust-lang.org/reference/types/impl-trait.html#r-type.impl-trait.intro

If those tests fail, please put them in a separate test file, and add a FIXME to the part of the code that fails. I can help with finding that part of the code once I see the error message.

Or if you know how to add support, you can just fix the code.

Failing Test

We can't test dyn Tuple because the Tuple trait isn't dyn-compatible, but we can test:

fn f<T>(#[splat] t: &dyn AsRef<T>) where T: Tuple {}

I'm not sure if we even want to support dyn or AsRef<Tuple>, but having failing tests like this helps us work out if we ever accidentally add support. And it makes it clear we don't expect it to work right now.

[rust-bors]

☔ The latest upstream changes (presumably #156816) made this pull request unmergeable. Please resolve the merge conflicts.