feat(divan): export benchmark tracing spans over OTLP behind an otel feature

[lvaroqui]

Add an opt-in otel feature to codspeed and codspeed-divan-compat that exports the tracing spans and events emitted by benchmarks over OTLP/HTTP.

With the feature enabled, divan::main() installs a global tracing subscriber when CODSPEED_OTEL_ENDPOINT is set, and flushes the spans when the run ends. CODSPEED_OTEL_FILTER takes EnvFilter directives to select what is exported, and defaults to info. Nothing changes when the variable is unset or the feature is off; all the OpenTelemetry dependencies are optional.

Benchmarks that already set their own global subscriber are left untouched. They can add divan::otel::layer() to it instead, as documented in the README.

Worth a careful look:

• Spans emitted by the exporter's own HTTP stack (hyper, reqwest, opentelemetry_*, …) are always filtered out. Exporting them would create more spans, in an endless loop.
• The OpenTelemetry service name is the bench executable name, with the hash cargo appends to it stripped.

benches/otel_example.rs emits nested spans and events, and a debug span that the default filter drops. Point it at any OTLP/HTTP receiver to check the output:

CODSPEED_OTEL_ENDPOINT=http://localhost:4318 cargo bench -p codspeed-divan-compat --features otel --bench otel_example

Closes COD-3862

[codspeed]

Merging this PR will regress 18 benchmarks

⚠️ 59 benchmarks measured no execution time

Nothing ran under measurement, usually because the compiler removed the code under test. These results are not comparable, so they count as unchanged.

Preventing compiler optimizations

• tsc (aarch64)
• iter
• iter_with_large_setup
• iter_with_large_drop
• iter_batched_small_input
• iter_batched_ref_small_input
• small_drop
• "*benchmark/" '
• with_input[5]
• with_input[5]
• …and 49 more

⚠️ 9 benchmarks spent significant time in system calls

System calls cannot be consistently instrumented, so they are not included in the measure, which understates the real cost. Please switch to the Walltime instrument to accurately measure system calls.

Measurement and system calls

• instant
• system_time
• recursive_memoized[HashMap<u64, u64>, 0]
• fibo_recursive_cached
• print_env_hello
• allocate
• bench_large_input_local
• bench_large_input
• iter_manual_simple

ℹ️ 49 benchmarks have no flame graph

• The flamegraph could not be generated because too much time was spent in runtime internals, making the visualization not useful for performance analysis. (1 benchmark)
• The flamegraph could not be generated because no execution data was recorded when the benchmark was run. Ensure that the benchmark has not been optimized out. (48 benchmarks)

• fib_in_thread_bench_local[30]
• iter_manual_async
• with_input[5]
• small_drop
• large_setup
• small_setup
• iter_with_setup
• "*benchmark/" '
• iter
• iter_with_setup
• …and 39 more

⚠️ 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
❌ 18 regressed benchmarks
✅ 558 untouched benchmarks

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
❌	Simulation	find_highest_set_bit[0]	172.3 ns	282 ns	-38.91%
❌	WallTime	graph_coloring[5]	1.1 µs	1.4 µs	-25.65%
❌	WallTime	hamiltonian_cycle[6]	1.1 µs	1.5 µs	-25.41%
❌	Simulation	rem	178.1 ns	232.3 ns	-23.32%
❌	Simulation	div	178.1 ns	232.3 ns	-23.32%
❌	WallTime	recursive[5]	34 ns	44 ns	-22.73%
❌	Simulation	iterative[5]	234.1 ns	289.6 ns	-19.18%
❌	Simulation	find_highest_set_bit[42]	229.3 ns	283.5 ns	-19.11%
❌	Simulation	find_highest_set_bit[1024]	229.3 ns	283.5 ns	-19.11%
❌	Simulation	find_highest_set_bit[255]	229.3 ns	283.5 ns	-19.11%
❌	Simulation	find_highest_set_bit[65535]	229.3 ns	283.5 ns	-19.11%
❌	Simulation	iterative[10]	245.4 ns	300.9 ns	-18.46%
❌	Simulation	iterative[20]	267.9 ns	323.5 ns	-17.17%
❌	Simulation	iterative[30]	290.4 ns	346 ns	-16.06%
❌	WallTime	sort	118.3 µs	139.2 µs	-15.02%
❌	Simulation	process_items_with_counter	1.8 µs	1.9 µs	-5.67%
❌	WallTime	add_two_integers[(255, 255)]	19 ns	20 ns	-5%
❌	WallTime	add_two_integers[(42, 13)]	19 ns	20 ns	-5%
⚡	WallTime	bench_array1[4]	2 ns	1 ns	×2
⚡	WallTime	iter_batched_ref_large_input	13 ns	7 ns	+85.71%
...	...	...	...	...	...

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-3862-featrust-otel-cargo-feature-in-codspeed-divan-compat (6c3f81a) with main (5aaf570)}