fix: incorrect link of flamegraph and URI

[not-matthias]

We modified the logic to exclude the warmup, which caused this regression. We can only have 1 pair of SampleStart/Stop markers (generated by measurement_start/stop) for one benchmark.

Instead of moving the measurement_start/stop functions around to exclude the warmup, we now pass a flag that prevents adding the BenchmarkStart/Stop markers.

[Copilot]

Pull request overview

This PR fixes the marker emission pattern for CodSpeed benchmarking by ensuring sample markers wrap around all benchmark repetitions, with benchmark start only being called once on the first repetition. Previously, start_benchmark was called on every repetition, which would emit multiple nested MARKER_TYPE_SAMPLE_START/END pairs. Now the markers are emitted in the correct format where a single sample start/end pair wraps around all benchmark iterations, with multiple benchmark start/stop markers within.

Changes:

• Added IsFirstRepetition() helper method to check if a benchmark runner is on its first iteration
• Modified start_benchmark call to only execute on the first repetition
• Added MARKER_TYPE_SAMPLE_START marker emission at the beginning of start_benchmark()
• Added MARKER_TYPE_SAMPLE_END marker emission at the beginning of end_benchmark()

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated no comments.

File	Description
google_benchmark/src/benchmark_runner.h	Added IsFirstRepetition() method to check if benchmark is on first repetition
google_benchmark/src/benchmark.cc	Modified condition to only call start_benchmark on first repetition
core/src/codspeed.cpp	Added sample start/end markers around benchmark execution

---

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

[codspeed-hq]

Merging this PR will degrade performance by 32.51%

⚡ 20 improved benchmarks
❌ 16 regressed benchmarks
✅ 156 untouched benchmarks

⚠️ Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
⚡	Simulation	DoubleTest[MyTemplatedFixture, double]	90.8 ns	61.7 ns	+47.3%
⚡	Simulation	BarTest[MyFixture]	90.8 ns	61.7 ns	+47.3%
❌	Simulation	BM_Template1_Capture[int_string_test]	60.6 ns	89.7 ns	-32.51%
❌	Simulation	BM_Capture_int[int_test]	61.1 ns	90.3 ns	-32.31%
⚡	Simulation	BM_Template2[int, double]	91.1 ns	61.9 ns	+47.09%
❌	Simulation	BM_Template1_Capture[two_type_test, int, double]	60.8 ns	90 ns	-32.41%
⚡	Simulation	FooTest[MyFixture]	90.8 ns	61.7 ns	+47.3%
⚡	Simulation	TestA[MyTemplate1, int]	90.8 ns	61.7 ns	+47.3%
⚡	Simulation	TestB[MyTemplate2, int, double]	90.8 ns	61.7 ns	+47.3%
❌	Simulation	BM_custom_args[1000]	90.8 ns	120 ns	-24.31%
⚡	Simulation	IntTest[MyTemplatedFixture, int]	90.8 ns	61.7 ns	+47.3%
⚡	Simulation	BM_Template1[int]	90.8 ns	61.7 ns	+47.3%
❌	Simulation	BM_Capture[int_test]	61.1 ns	90.3 ns	-32.31%
❌	Simulation	BM_Capture[int_string_test]	60.6 ns	89.7 ns	-32.51%
❌	Simulation	BM_with_args[1000]	90.8 ns	120 ns	-24.31%
❌	Simulation	BM_custom_args[100]	90.8 ns	120 ns	-24.31%
❌	Simulation	BM_with_args[100]	90.8 ns	120 ns	-24.31%
⚡	Simulation	BM_rand_vector	151.4 ns	122.2 ns	+23.86%
❌	Simulation	BM_custom_name_in_namespace	61.7 ns	90.8 ns	-32.11%
❌	Simulation	BM_with_multiple_args[100/200]	153.3 ns	182.5 ns	-15.98%
...	...	...	...	...	...

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

---

_{Comparing cod-2022-incorrect-link-of-uri-and-flamegraph (660b1f9) with main (56e942a)}

[GuillaumeLagrange]

https://codspeed.io/CodSpeedHQ/codspeed-cpp/branches/cod-2022-incorrect-link-of-uri-and-flamegraph?utm_source=github&utm_medium=comment-v2&utm_content=benchmark&uri=examples%2Fgoogle_benchmark_cmake%2Fpause_timing_bench.hpp%3A%3ABM_large_setup&runnerMode=Instrumentation&sectionId=benchmark-comparison-section-comparison-passed

Feels like there are still some issues.

According to the code, we should never see the expensive_operation w/ fibo here

[not-matthias]

codspeed.io/CodSpeedHQ/codspeed-cpp/branches/cod-2022-incorrect-link-of-uri-and-flamegraph?utm_source=github&utm_medium=comment-v2&utm_content=benchmark&uri=examples%2Fgoogle_benchmark_cmake%2Fpause_timing_bench.hpp%3A%3ABM_large_setup&runnerMode=Instrumentation&sectionId=benchmark-comparison-section-comparison-passed

Feels like there are still some issues.

According to the code, we should never see the expensive_operation w/ fibo here

This seems to already have been the case 2 months ago: https://codspeed.io/CodSpeedHQ/codspeed-cpp/runs/692085eb175fc14ee5058b33?uri=examples%2Fgoogle_benchmark_cmake%2Fpause_timing_bench.hpp%3A%3ABM_large_setup&runnerMode=Instrumentation&sectionId=benchmark-comparison-section-single-run-benchmarks

I'll create a new ticket and fix it there