Documentation
¶
Overview ¶
Package benchcmp compares Go benchmark output without applying pass/fail thresholds.
Index ¶
- func Median(values []float64) (float64, error)
- func ValidateCompleteness(results Results, expectedBenchmarks int, expectedSamples int) error
- func ValidateExpectedNames(results Results, expected []string) error
- func WriteMarkdown(w io.Writer, baseline Results, candidate Results, baselineLabel string, ...) error
- type Distribution
- type Metric
- type Results
- type Samples
Constants ¶
This section is empty.
Variables ¶
This section is empty.
Functions ¶
func ValidateCompleteness ¶
ValidateCompleteness checks that a benchmark run contains the expected number of benchmarks and observations for every reported metric. A zero expectation disables that dimension.
func ValidateExpectedNames ¶
ValidateExpectedNames checks that results contain exactly the normalized benchmark names in expected. An empty expected manifest disables the check.
func WriteMarkdown ¶
func WriteMarkdown( w io.Writer, baseline Results, candidate Results, baselineLabel string, candidateLabel string, ) error
WriteMarkdown writes a deterministic comparison table. Ratios are candidate divided by baseline; the function deliberately makes no threshold or pass/fail decision.
Types ¶
type Distribution ¶
Distribution summarizes one metric without hiding its observation count or middle-half spread.
func Summarize ¶
func Summarize(values []float64) (Distribution, error)
Summarize returns the sample count, median, and linearly interpolated first and third quartiles without changing the input order.
type Metric ¶
type Metric string
Metric identifies one of the standard Go benchmark metrics compared by this package.
Source Files