Syntactically reject leading parenthesized precise capturing lists in bare trait object types ((use<…>)+)

[fmease]

Follow-up to #162269. Addresses fmease/rasur#7 (item 6).

_{(No LLM was or will be used by me during the entire creation process of this PR)}

[rustbot]

r? @adwinwhite

rustbot has assigned @adwinwhite.
They will have a look at your PR within the next two weeks and either review your PR or reassign to another reviewer.

Use r? to explicitly pick a reviewer

Why was this reviewer chosen?

The reviewer was selected based on:

• Owners of files modified in this PR: compiler, parser
• compiler, parser expanded to 76 candidates
• Random selection from 21 candidates

[fmease]

This is so niche that it does not need a crater run IMO and it might not even require an FCP.

[fmease]

Context for T-lang:

We generally only support parenthesizing trait bounds, not however any other kind of bound (namely, outlives-bounds and use-bounds) and as it stands there would be no use in generalizing this part of the grammar. So we allow bounds like (Trait), (for<'a> path::to::Trait<'a>) and (Fn() -> i32) but we don't allow ('a) or (use<N>)...

...Well, when precise capturing lists were introduced & stabilized, something was overlooked: While we correctly reject bare trait object types¹ like Trait + (use<T>) (where the use-bound doesn't come first) before this PR we would incorrectly syntactically accept bare trait object types like (use<T>) + Trait or (use<>)+ (where the use-bounds come first).

Note that all of this only concerns the grammar of Rust as use-bounds are semantically invalid in trait object types anyway. This is just part of my ongoing efforts to deep clean Rust's grammar + rustc's parser.

Arguably this is super niche and thus I actually think that merging this w/o a T-lang FCP would be fine but I wanted to at least inform you of this.

The Reference doesn't need updating as it already conforms to what I expect & propose here. See also TraitObjectType, Bound etc.

Footnotes

1. Which are syntactically legal in all editions by the way; they're merely semantically illegal in Rust >=2021. ↩

[tmandry]

@rfcbot fcp merge lang

[rust-rfcbot]

@tmandry has proposed to merge this. The next step is review by the rest of the tagged team members:

• @joshtriplett
• @nikomatsakis
• @scottmcm
• @tmandry
• @traviscross

No concerns currently listed.

Once a majority of reviewers approve (and at most 2 approvals are outstanding), this will enter its final comment period. If you spot a major issue that hasn't been raised at any point in this process, please speak up!

cc @rust-lang/lang-advisors: FCP proposed for lang, please feel free to register concerns.
See this document for info about what commands tagged team members can give me.

[traviscross]

Thanks @fmease.

@rfcbot reviewed