fix: Filter suggestion parts that match existing code

[Muscraft]

While testing my changes to make rustc use annotate-snippets, I encountered a new clippy test failure stemming from two suggestion output changes in #145273. The new output in these two cases feels like a regression as it is not as clear as the old output, and adds unnecessary information.

Before #145273 (Diff style)

After #145273 ("multi-line" style)

The reason for the change was that a new suggestion part (which matches existing code) was added on a different line than the existing parts, causing the suggestion style to change from Diff to "multi-line". Since this new part matches existing code, no code changes show up in the output for it, but it still makes the suggestion style "multi-line" when it doesn't need to be.

To get the old output back, I made it so that suggestion parts that perfectly match existing code get filtered out.

try-job: aarch64-apple

[rustbot]

r? @petrochenkov

rustbot has assigned @petrochenkov.
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

[rustbot]

Some changes occurred in src/tools/clippy

cc @rust-lang/clippy

[petrochenkov]

@bors r+

[bors]

📌 Commit f696cd8 has been approved by petrochenkov

It is now in the queue for this repository.

[jieyouxu]

This probably also fixes (or at least suppresses tests/crashes/131762.rs): #146139 (comment)

Could you convert it into a ui test or remove if redundant?

@bors r-

[jieyouxu]

(Let me double-check)

@bors try jobs=aarch64-apple

[Muscraft]

I have been unable to reproduce the test failure on my Mac locally, so I might have to use try builds to figure it out.

@bors try jobs=aarch64-apple

[rustbot]

This PR was rebased onto a different master 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.

[Muscraft]

I spent a ton of time looking into why the ICE was no longer happening on CI, but would still happen locally.

Before my change, something causes one of the spans in tests/crashes/131762.rs to be malformed, which causes a suggestion to have overlapping parts, one of which is empty (span is empty and snippet is empty). When running in CI (or debug-assertions weren't enabled), the overlapping Parts would fail this assert!, causing the expected ICE. When running with debug-assertions enabled, the empty Part would fail this debug_assert_eq!, causing an ICE before the other assert! could be reached. In both cases, tests/crashes/131762.rs would ICE, so the test would pass. With my original changes, when debug-assertions were enabled, it would ICE on the debug_assert_eq!, so the test would pass. However, when debug-assertions were disabled, my changes would filter out the empty Part before the assert!, making it so there were no overlapping Parts, so no ICE happened.

After thinking about it for a bit, I realized that my changes were only masking the ICE, not fixing the cause of it. To make it so, I was no longer "masking" the ICE; I moved the assert! checking for disjoint spans before the filter for suggestion parts that match existing code, making it so an ICE would still happen if a suggestion's spans weren't disjoint.

@rustbot review

[petrochenkov]

@bors r+

[bors]

📌 Commit b307a11 has been approved by petrochenkov

It is now in the queue for this repository.

[bors]

⌛ Testing commit b307a11 with merge 99317ef...

[bors]

☀️ Test successful - checks-actions
Approved by: petrochenkov
Pushing 99317ef to master...

[github-actions]

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 9cd272d (parent) -> 99317ef (this PR)

Test differences

Show 4 test diffs

4 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 99317ef14d0be42fa4039eea7c5ce50cb4e9aee7 --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

1. dist-aarch64-linux: 8950.7s -> 6408.0s (-28.4%)
2. dist-various-1: 3687.6s -> 4222.4s (14.5%)
3. x86_64-rust-for-linux: 2552.0s -> 2864.5s (12.2%)
4. x86_64-gnu-stable: 7912.3s -> 7049.0s (-10.9%)
5. i686-gnu-nopt-1: 7251.2s -> 7998.3s (10.3%)
6. x86_64-gnu-llvm-19-3: 6371.6s -> 6969.0s (9.4%)
7. dist-apple-various: 4092.4s -> 4461.9s (9.0%)
8. aarch64-gnu-llvm-19-1: 3342.3s -> 3634.6s (8.7%)
9. armhf-gnu: 5128.8s -> 4718.9s (-8.0%)
10. i686-gnu-2: 5430.2s -> 5859.7s (7.9%)

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]

Finished benchmarking commit (99317ef): comparison URL.

Overall result: no relevant changes - no action needed

@rustbot label: -perf-regression

Instruction count

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

Max RSS (memory usage)

Results (primary 1.6%, secondary -2.1%)

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

	mean	range	count
Regressions ❌ (primary)	3.9%	[3.9%, 3.9%]	1
Regressions ❌ (secondary)	2.4%	[2.4%, 2.4%]	1
Improvements ✅ (primary)	-0.7%	[-0.7%, -0.7%]	1
Improvements ✅ (secondary)	-3.2%	[-5.5%, -2.1%]	4
All ❌✅ (primary)	1.6%	[-0.7%, 3.9%]	2

Cycles

Results (primary 2.1%)

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

	mean	range	count
Regressions ❌ (primary)	2.1%	[2.1%, 2.1%]	1
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	2.1%	[2.1%, 2.1%]	1

Binary size

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

Bootstrap: 466.363s -> 468.222s (0.40%)
Artifact size: 390.44 MiB -> 390.57 MiB (0.03%)