Latency percentiles are currently hardcoded. Users running SLO-sensitive benchmarks may need p99.9 or custom percentiles.
What to add:
cli:
httpress -n 10000 https://example.com --percentiles 50,90,99,99.9
library:
Benchmark::builder()
.url("https://example.com")
.percentiles(vec![50.0, 90.0, 99.0, 99.9])
.build()?
.run()
.await?;
Relevant files:
src/cli.rs - add flag
src/config.rs - add --percentiles field with default of [50.0, 90.0, 95.0, 99.0]
src/benchmark.rs - add .percentiles() builder method
src/metrics.rs- make percentile computation dynamic
src/main.rs - update display
notes:
- make sure to also update the readme
cli useage / options section to include the new flag
- also add a documentation comment to
.percentiles() builder method with a description and a minimal example (you can look at the other functions for reference).
Latency percentiles are currently hardcoded. Users running SLO-sensitive benchmarks may need p99.9 or custom percentiles.
What to add:
cli:
library:
Relevant files:
src/cli.rs- add flagsrc/config.rs- add--percentilesfield with default of[50.0, 90.0, 95.0, 99.0]src/benchmark.rs- add.percentiles()builder methodsrc/metrics.rs- make percentile computation dynamicsrc/main.rs- update displaynotes:
cli useage/optionssection to include the new flag.percentiles()builder method with a description and a minimal example (you can look at the other functions for reference).