Simplify `HashStable` by nnethercote · Pull Request #155018 · rust-lang/rust

nnethercote · 2026-04-09T00:40:40Z

View all comments

This PR:

Simplifies the HashStable trait, by moving its generic parameter from the trait to its single method.
Eliminates the need for the non-obvious derive(HashStable)/derive(HashStable_Generic) distinction.
Reduces the need for, and clarifies, the non-obvious derive(HashStable)/derive(HashStable_NoContext) distinction.

r? @fee1-dead

nnethercote · 2026-04-09T01:01:42Z

Why on earth is Span 96 bits on aarch64-gnu-llvm-21-1 when it only contains a u32 and two u16s?

nnethercote · 2026-04-09T02:19:00Z

I just tried PR on an aarch64-unknown-linux-gnu box that I have and it compiled fine. Weird.

petrochenkov · 2026-04-09T11:52:08Z

+/// visible.
+pub trait HashStableContext {
+    /// The main event: stable hashing of a span.
+    fn span_hash_stable(&mut self, span: RawSpan, hasher: &mut StableHasher);


View changes since the review

Suggested change

fn span_hash_stable(&mut self, span: RawSpan, hasher: &mut StableHasher);

fn span_hash_stable(&mut self, span: Self::Span, hasher: &mut StableHasher);

Perhaps Span, DefId and RawDefPathHash could be associated types in HashStableContext?
Then impl HashStableContext for StableHashingContext could fill them with concrete types.

If that works then "type erased" versions of the aforementioned types could be avoided.

I tried that but I couldn't get it to work.

Specifically, if I add an associate type MySpan to the trait, then in Span::hash_stable I get this error:

error[E0308]: mismatched types --> compiler/rustc_span/src/lib.rs:2793:30 | 2793 | hcx.span_hash_stable(*self, hasher) | ---------------- ^^^^^ expected associated type, found `Span` | | | arguments to this method are incorrect | = note: expected associated type `<Hcx as HashStableContext>::MySpan` found struct `span_encoding::Span` help: consider constraining the associated type `<Hcx as HashStableContext>::MySpan` to `span_encoding::Span` | 2791 | fn hash_stable<Hcx: HashStableContext<MySpan = span_encoding::Span>>(&self, hcx: &mut Hcx, hasher: &mut StableHasher) { | ++++++++++++++++++++++++++++++

Then if I try the help suggestion I get this error:

error[E0271]: type mismatch resolving `<Hcx as HashStableContext>::MySpan == Span` --> compiler/rustc_span/src/lib.rs:2791:43 | 2791 | ...ashStableContext<MySpan = Span>>(&self, hcx: &mut Hcx, hasher: &mut StableHasher) { | ^^^^^^^^^^^^^ expected `Span`, found associated type | = note: expected struct `span_encoding::Span` found associated type `<Hcx as HashStableContext>::MySpan` note: the requirement `<Hcx as HashStableContext>::MySpan == span_encoding::Span` appears on the `impl`'s method `hash_stable` but not on the corresponding trait's method --> compiler/rustc_data_structures/src/stable_hasher.rs:82:8

In other words, the impl method no longer matches the trait method.

Maybe there's a way around this? I couldn't find one.

You can probably make it work with appropriate bounds on the associated types for the impl, but that's going to be crate dependent so still doesn't play nicely with the proc macro. I think you'd need to keep the generic on the trait too.

nnethercote · 2026-04-10T00:33:59Z

This is blocked by #154924.

nnethercote · 2026-04-10T00:34:37Z

@bors try @rust-timer queue

nnethercote · 2026-04-10T05:25:10Z

@bors try @rust-timer queue

Simplify `HashStable`

rust-bors · 2026-04-10T07:33:52Z

☀️ Try build successful (CI)
Build commit: 959be73 (959be73237e12c458b274d84f125d94f71fe1d9e, parent: dd82fd2034e5c86c6f0f8070dc5679703cc4fe03)

rust-timer · 2026-04-10T08:14:19Z

Finished benchmarking commit (959be73): comparison URL.

Overall result: ❌ regressions - please read:

Benchmarking means the PR may be perf-sensitive. It's automatically marked not fit for rolling up. Overriding is possible but disadvised: it risks changing compiler perf.

Next, please: If you can, justify the regressions found in this try perf run in writing along with @rustbot label: +perf-regression-triaged. If not, fix the regressions and do another perf run. Neutral or positive results will clear the label automatically.

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

Instruction count

Our most reliable metric. Used to determine the overall result above. However, even this metric can be noisy.

	mean	range	count
Regressions ❌ (primary)	0.4%	[0.1%, 1.2%]	68
Regressions ❌ (secondary)	0.4%	[0.2%, 1.5%]	23
Improvements ✅ (primary)	-0.1%	[-0.2%, -0.1%]	3
Improvements ✅ (secondary)	-0.2%	[-0.3%, -0.1%]	4
All ❌✅ (primary)	0.4%	[-0.2%, 1.2%]	71

Max RSS (memory usage)

Results (secondary 6.0%)

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)	6.0%	[6.0%, 6.0%]	1
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-	-	0

Cycles

Results (secondary 2.8%)

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)	2.8%	[2.8%, 2.8%]	1
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-	-	0

Binary size

This perf run didn't have relevant results for this metric.

Bootstrap: 488.849s -> 492.643s (0.78%)
Artifact size: 395.54 MiB -> 395.57 MiB (0.01%)

rustbot · 2026-04-29T22:45:07Z

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.

nnethercote · 2026-04-29T22:45:33Z

I added a new commit with the requested comment.

@bors r=fee1-dead p=1 (conflict-prone)

rust-bors · 2026-04-29T22:45:37Z

📌 Commit 20060c6 has been approved by fee1-dead

It is now in the queue for this repository.

@fee1-dead

Simplify `HashStable` This PR: - Simplifies the `HashStable` trait, by moving its generic parameter from the trait to its single method. - Eliminates the need for the non-obvious `derive(HashStable)`/`derive(HashStable_Generic)` distinction. - Reduces the need for, and clarifies, the non-obvious `derive(HashStable)`/`derive(HashStable_NoContext)` distinction. r? @fee1-dead

rust-log-analyzer · 2026-04-30T04:08:51Z

The job dist-x86_64-msvc failed! Check out the build log: (web) (plain enhanced) (plain)

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

rust-bors · 2026-04-30T04:13:54Z

💔 Test for e4a01d5 failed: CI. Failed job:

auto - dist-x86_64-msvc (web logs, enhanced plaintext logs)

nnethercote · 2026-04-30T04:18:59Z

Unclear what went wrong. The last lines in the failing log are "Vendoring" steps. Doesn't seem related to anything in this PR.

@bors retry

rust-bors · 2026-04-30T10:38:50Z

☀️ Test successful - CI
Approved by: fee1-dead
Duration: 3h 7m 54s
Pushing f53b654 to main...

github-actions · 2026-04-30T10:42:23Z

What is this?

This is an experimental post-merge analysis report that shows differences in test outcomes between the merged PR and its parent PR.

Comparing 57f772f (parent) -> f53b654 (this PR)

Test differences

Show 80 test diffs

80 doctest diffs were found. These are ignored, as they are noisy.

Test dashboard

Run

cargo run --manifest-path src/ci/citool/Cargo.toml -- \
    test-dashboard f53b654a8882fd5fc036c4ca7a4ff41ce32497a6 --output-dir test-dashboard

And then open test-dashboard/index.html in your browser to see an overview of all executed tests.

Job duration changes

x86_64-gnu-llvm-21: 1h 39m -> 1h 11m (-28.1%)
i686-msvc-1: 2h 16m -> 2h 54m (+27.7%)
dist-aarch64-linux: 2h 30m -> 1h 54m (-24.3%)
x86_64-gnu-gcc: 1h 24m -> 1h 5m (-22.9%)
x86_64-gnu-llvm-22-3: 1h 55m -> 1h 29m (-22.7%)
x86_64-rust-for-linux: 39m 2s -> 46m 12s (+18.4%)
dist-x86_64-solaris: 1h 29m -> 1h 45m (+18.0%)
i686-gnu-nopt-1: 1h 49m -> 2h 7m (+16.3%)
dist-apple-various: 1h 40m -> 1h 56m (+16.0%)
dist-x86_64-apple: 1h 50m -> 2h 6m (+14.8%)

How to interpret the job duration changes?

Job durations can vary a lot, based on the actual runner instance
that executed the job, system noise, invalidated caches, etc. The table above is provided
mostly for t-infra members, for simpler debugging of potential CI slow-downs.

rust-timer · 2026-04-30T11:19:17Z

Finished benchmarking commit (f53b654): comparison URL.

Overall result: ❌ regressions - no action needed

@rustbot label: -perf-regression

Instruction count

Our most reliable metric. Used to determine the overall result above. However, even this metric can be noisy.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	0.3%	[0.2%, 0.3%]	4
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-	-	0

Max RSS (memory usage)

Results (primary -0.9%, secondary -3.0%)

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)	-0.9%	[-0.9%, -0.9%]	1
Improvements ✅ (secondary)	-3.0%	[-3.5%, -2.0%]	4
All ❌✅ (primary)	-0.9%	[-0.9%, -0.9%]	1

Cycles

This perf run didn't have relevant results for this metric.

Binary size

Results (secondary -0.1%)

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)	-0.1%	[-0.1%, -0.1%]	1
All ❌✅ (primary)	-	-	0

Bootstrap: 483.007s -> 487.787s (0.99%)
Artifact size: 391.00 MiB -> 390.97 MiB (-0.01%)