Set up API to make it possible to pass closures instead of AttributeLint

[GuillaumeGomez]

View all comments

Part of #153099.

This PR sets up the base implementations needed to remove AttributeLintKind entirely and migrate two variants as examples.

r? @JonathanBrouwer

[rustbot]

Some changes occurred in compiler/rustc_attr_parsing

cc @jdonszelmann, @JonathanBrouwer

These commits modify the Cargo.lock file. Unintentional changes to Cargo.lock can be introduced when switching branches and rebasing PRs.

If this was unintentional then you should revert the changes before this PR is merged.
Otherwise, you can ignore this comment.

[rustbot]

JonathanBrouwer is currently at their maximum review capacity.
They may take a while to respond.

[GuillaumeGomez]

Added the missing file and fixed the typo. Time to run a perf check.

@bors try @rust-timer queue

[rust-bors]

☀️ Try build successful (CI)
Build commit: 8dcd0b9 (8dcd0b95692c85cfd497ecfb5770e55e4796ba80, parent: 23903d01c237d7c7d4fb62b82ca846bc45de4e0c)

[rust-timer]

Finished benchmarking commit (8dcd0b9): comparison URL.

Overall result: no relevant changes - no action needed

Benchmarking this pull request means it may be perf-sensitive – we'll automatically label it not fit for rolling up. You can override this, but we strongly advise not to, due to possible changes in compiler perf.

@bors rollup=never
@rustbot label: -S-waiting-on-perf -perf-regression

Instruction count

This benchmark run did not return any relevant results for this metric.

Max RSS (memory usage)

Results (primary 3.6%, secondary 6.4%)

A less reliable metric. May be of interest, but not used to determine the overall result above.

	mean	range	count
Regressions ❌ (primary)	3.6%	[3.6%, 3.6%]	1
Regressions ❌ (secondary)	6.4%	[6.4%, 6.4%]	1
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	3.6%	[3.6%, 3.6%]	1

Cycles

Results (secondary -2.0%)

A less reliable metric. May be of interest, but not used to determine the overall result above.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-2.0%	[-2.0%, -2.0%]	1
All ❌✅ (primary)	-	-	0

Binary size

This benchmark run did not return any relevant results for this metric.

Bootstrap: 494.635s -> 485.284s (-1.89%)
Artifact size: 395.08 MiB -> 397.09 MiB (0.51%)

[GuillaumeGomez]

Nice. =D

[JonathanBrouwer]

Is this correct? If the lint/id/span of the lint stays the same, but something in the labels changes, then the hash of the DynAttribute would remain unchanged

[GuillaumeGomez]

I don't see a way around that, whether it's dyn Diagnostic or a callback. Although, do you see a case where a same AttributeLint is emitted on both the same HirId and span but with a different error?

[JonathanBrouwer]

I don't really like that a closure is needed here, I experimented a bit locally and this seems to be because we store the result in a query, and to call IntoDiagnostic we need ownership of this thing, which the query can't provide.

We could theoretically work around that with Steal but that seems a bit too extreme tool for such a small annoyance

[JonathanBrouwer]

Maybe taking a diagnostic and requiring it to be Clone solves it?

[GuillaumeGomez]

To be checked but it's an idea I had in mind, just didn't try it out yet.

[GuillaumeGomez]

After testing out a bit, I think the current approach is the best one we can have because Box<dyn Diagnostic> is unhappy because dyn Diagnostic is not Sized (obviously) and keeping a &dyn Diagnostic creates more issues than it solves. I think it's (sadly) the best "form" we can have for now.

[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.

[GuillaumeGomez]

Fixed merge conflicts.