Implement derives for Reborrow and CoerceShared

[P8L1]

This adds built-in derive support for the experimental Reborrow and CoerceShared marker traits behind #![feature(reborrow)].

The derives generate the same marker impls users would write manually, while preserving the existing coherence and structural validation path for the experimental reborrow traits.

Please note this impl:

• Preserves generics, lifetimes, const params, and where clauses in generated impls.
• Preserves multiple lifetime parameters in generated impls instead of duplicating the current experimental one-lifetime limitation in the derive layer.
• Requires an explicit CoerceShared target type through the derive helper attribute.
• Adds UI coverage for valid derives, feature gating, malformed helper attributes, unsupported item kinds, validation failures, and multiple-lifetime derive behavior.
• Makes the derive layer future-proof for multiple lifetimes: it preserves all lifetime parameters and generates the same impl shape as a manual impl.

@rustbot label F-reborrow
Tracking: #145612

[rustbot]

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

[rustbot]

r? @tiif

rustbot has assigned @tiif.
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
• compiler expanded to 73 candidates
• Random selection from 18 candidates

[ChayimFriedman2]

Please separate the rust-analyzer changes to a new PR in the rust-analyzer repo, there is no reason for them to be here.

View changes since this review

[rustbot]

Reminder, once the PR becomes ready for a review, use @rustbot ready.

[P8L1]

@rustbot ready

[aapoalas]

suggestion: I would've gone with derive(CoerceShared(Target)).

View changes since the review

[oli-obk]

those are hard to support in the current derive infra.

[oli-obk]

Should def be an open question on the tracking issue, as #[coerce_shared(Thing<'a, T>)] looks a bit unclear to me, maybe #[coerce_shared_target = Thing<'a, T>] if our attribute syntax permits that?

[P8L1]

Ok, I checked the available attribute forms. The current grammar does not accept '#[coerce_shared_target = Thing<'a, T>], since the right-hand side of an = attribute is parsed as an expression and stored as AttrArgs::Eq { expr: Box<Expr>, .. }, while Thing<'a, T> is type syntax and fails expression parsing.

Also, since the current derive pipeline diagnoses arguments on derive paths and resolves only the trait path, we currently reject derive(CoerceShared(Target))

A structured helper such as #[coerce_shared(target = Thing<'a, T>)] is technically feasible because delimited attributes retain their token stream, allowing the builtin derive to parse target = followed by parse_ty().

[aapoalas]

Thank you so much for this work. I'm really thankful for the help; I expect this would've taken me quite a bit of time to figure out! <3

View changes since this review

[tiif]

I don't have enough context to review this, so I will leave this to the compiler champion.

r? @oli-obk

[P8L1]

Hi @oli-obk, no rush, but I wanted to check that this PR is still on your radar.

I know about the usual two-week review window; I’m just making sure it did not get missed after the latest updates. Thanks!

[P8L1]

Oli seems to be very busy ATM so thus:

r? @dingxiangfei2009

Cc. @aapoalas

[P8L1]

@dingxiangfei2009 It's been a while, just a reminder to review. No pressure thought!

[oli-obk]

Good thing I didn't unsubscribe, even if it was reassigned from me. I was a bit busy catching up in May as I was not doing much Rust stuff before it.

Generally lgtm, slightly worried about spans without expansion info causing funny diagnostics, but we'll add that when necessary (or testable)

View changes since this review

[oli-obk]

those are hard to support in the current derive infra.

[oli-obk]

Should def be an open question on the tracking issue, as #[coerce_shared(Thing<'a, T>)] looks a bit unclear to me, maybe #[coerce_shared_target = Thing<'a, T>] if our attribute syntax permits that?

[rustbot]

This PR was rebased onto a different main commit. Here's a range-diff highlighting what actually changed.

Rebasing is a normal part of keeping PRs up to date, so no action is needed—this note is just to help reviewers.

[P8L1]

@rustbot ready