fix(criterion-compat): fix URI formatting when criterion_group!/criterion_main! are bypassed

[not-matthias]

No description provided.

[Copilot]

Pull request overview

Fixes CodSpeed Criterion-compat benchmark URI construction when users bypass criterion_group! / criterion_main! (e.g., custom main), ensuring URIs don’t include empty segments and that current_file is derived from the callsite when missing.

Changes:

• Build benchmark URIs from non-empty components (file / macro group / group name / function) to avoid malformed :: sequences.
• Derive current_file from #[track_caller] callsite when criterion_group! didn’t set it.
• Add a new bench reproducer for the “custom main” scenario and register it in Cargo.toml.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 3 comments.

File	Description
crates/criterion_compat/src/compat/group.rs	Makes URI formatting robust to missing file and/or macro group components.
crates/criterion_compat/src/compat/criterion.rs	Adds callsite-based fallback to populate current_file when macros aren’t used.
crates/criterion_compat/benches/repro_custom_main.rs	Adds a reproducer bench for bypassing criterion_group!/criterion_main!.
crates/criterion_compat/Cargo.toml	Registers the new repro_custom_main bench target.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

You can also share your feedback on Copilot code review. Take the survey.

[codspeed-hq]

Merging this PR will not alter performance

⚠️ Different runtime environments detected

Some benchmarks with significant performance changes were compared across different runtime environments,
which may affect the accuracy of the results.

Open the report in CodSpeed to investigate

⚡ 18 improved benchmarks
❌ 16 regressed benchmarks
✅ 549 untouched benchmarks
🆕 8 new benchmarks

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
❌	WallTime	find_highest_set_bit[255]	5 ns	47 ns	-89.36%
❌	WallTime	bench_array2[1]	1 ns	2 ns	-50%
❌	Simulation	fibo_10	90 ns	119.2 ns	-24.48%
❌	Simulation	add	91.4 ns	120.6 ns	-24.19%
❌	Simulation	mul	91.4 ns	120.6 ns	-24.19%
❌	Simulation	find_highest_set_bit[42]	122.2 ns	151.4 ns	-19.27%
❌	Simulation	find_highest_set_bit[1024]	122.2 ns	151.4 ns	-19.27%
❌	Simulation	find_highest_set_bit[255]	122.2 ns	151.4 ns	-19.27%
❌	Simulation	find_highest_set_bit[65535]	122.2 ns	151.4 ns	-19.27%
❌	WallTime	iter_batched_per_iteration	46 ns	56 ns	-17.86%
❌	WallTime	iter_batched_ref_large_input	5 ns	6 ns	-16.67%
❌	WallTime	iter_with_setup	45 ns	54 ns	-16.67%
❌	WallTime	iter_batched_large_input	7 ns	8 ns	-12.5%
❌	Simulation	iter_manual_simple	536.9 ns	566.1 ns	-5.15%
❌	Simulation	uppercase_chars_with_counter	2.1 µs	2.2 µs	-4.03%
❌	Simulation	process_items_with_counter	835.8 ns	865 ns	-3.37%
⚡	WallTime	hamiltonian_cycle[5]	1,439 ns	902 ns	+59.53%
⚡	WallTime	vec_copy_with_bytes_counter	57 ns	36 ns	+58.33%
⚡	WallTime	count_set_bits[0]	11 ns	7 ns	+57.14%
⚡	WallTime	string_processing_multi_counter	574 ns	370 ns	+55.14%
...	...	...	...	...	...

ℹ️ Only the first 20 benchmarks are displayed. Go to the app to view all benchmarks.

Tip

Investigate this regression by commenting @codspeedbot fix this regression on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing cod-2324-investigate-issues-with-uri-formatting-in-codspeed-criterion (ec46807) with main (99c7b5a)}

[GuillaumeLagrange]

olgtm

[GuillaumeLagrange]

@not-matthias https://codspeed.io/CodSpeedHQ/codspeed-rust/branches/cod-2324-investigate-issues-with-uri-formatting-in-codspeed-criterion?uri=%3A%3A%3A%3Amy_group%3A%3Amy_bench&runnerMode=WallTime&utm_source=github&utm_medium=comment-v2&utm_content=benchmark&page=1

Is this URI expected though ?

[not-matthias]

@not-matthias codspeed.io/CodSpeedHQ/codspeed-rust/branches/cod-2324-investigate-issues-with-uri-formatting-in-codspeed-criterion?uri=%3A%3A%3A%3Amy_group%3A%3Amy_bench&runnerMode=WallTime&utm_source=github&utm_medium=comment-v2&utm_content=benchmark&page=1

Is this URI expected though ?

oh wait, that's the one i wanted to fix. i'll have a look

[greptile-apps]

Greptile Summary

Fixes malformed URIs like ::::my_group::my_bench that occurred when users bypassed criterion_group!/criterion_main! macros, leaving current_file and macro_group empty. The fix has two parts: build_uri replaces bare format! URI construction by filtering out empty segments before joining with ::, and ensure_current_file lazily resolves the benchmark file path from the #[track_caller] call site when set_current_file was never called.

• build_uri + caller_file_path are added to crates/codspeed/src/utils.rs as shared utilities, with unit tests covering all three segment-population scenarios (all empty, partial, full).
• ensure_current_file is added to both the compat Criterion and the fork Criterion, and called unconditionally from benchmark_group; it is a no-op when set_current_file has already been invoked by the macro-generated code.
• All public bench-entry-point methods (bench_function, bench_with_input, benchmark_group) in both code paths gain #[track_caller] so the implicit caller location propagates correctly through the chain to Location::caller().

Confidence Score: 5/5

Safe to merge — the change is a targeted bug fix with no breaking API surface and a clearly scoped impact limited to URI generation when the standard macros are bypassed.

Both code paths (compat layer and criterion fork) are updated consistently, the fix is guarded by an idempotent early-return in ensure_current_file, and the new utility functions are exercised by unit tests and an integration benchmark covering the exact bypass patterns described in the issue.

No files require special attention.

Important Files Changed

Filename	Overview
crates/codspeed/src/utils.rs	Adds build_uri (filters empty :: segments) and caller_file_path (resolves call site via #[track_caller]), with accompanying tests.
crates/criterion_compat/src/compat/criterion.rs	Adds ensure_current_file helper and #[track_caller] on public bench methods; calls ensure_current_file from benchmark_group for lazy file-path resolution.
crates/criterion_compat/src/compat/group.rs	Switches URI construction from bare format! to build_uri, eliminating empty :: segments when macro_group or file path is unset.
crates/criterion_compat/criterion_fork/src/lib.rs	Adds ensure_current_file in the codspeed sub-module, #[track_caller] on bench entry points, and calls ensure_current_file from benchmark_group.
crates/criterion_compat/criterion_fork/src/analysis/mod.rs	Replaces inline format! URI construction with build_uri, consistent with the compat layer's group.rs change.
crates/criterion_compat/benches/custom_main.rs	New integration benchmark demonstrating both bypass patterns (direct call and criterion_group!-routed call) to validate correct URI generation.
crates/criterion_compat/Cargo.toml	Registers the new custom_main bench target with harness = false.

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A["User benchmark file\n(calls bench_function / benchmark_group)"]

    subgraph PatternA ["Pattern A — bypass criterion_group!"]
        A -->|"#[track_caller] chain"| BG["benchmark_group()"]
        BG --> ECF["ensure_current_file()"]
        ECF --> CFP["caller_file_path()\nLocation::caller() → user file path"]
    end

    subgraph PatternB ["Pattern B — through criterion_group! macro"]
        A2["criterion_group!-generated fn"] -->|"set_current_file()\nset_macro_group()"| BG2["benchmark_group()"]
        BG2 --> ECF2["ensure_current_file()\n(no-op — already set)"]
    end

    CFP --> URI["build_uri(&[file, macro_group, group_name])\nfilters empty segments"]
    ECF2 --> URI

    URI --> OUT["crates/...bench.rs::group::bench\n(no :::::: artifacts)"]

Loading

_{Reviews (3): Last reviewed commit: "test(criterion_compat): add custom main ..." | Re-trigger Greptile}

[not-matthias]

Works now!

[GuillaumeLagrange]

olgtm