HTTP Code Check & Benchmark
A simple trick to make your code look swag with an introduction to benchmark testing.
I’ll show you a code trick to check if an HTTP status is OK (between 200 and 299).
It will also allow me to introduce the benchmark in Go.
Example
code.go:
package main
func Code(n int) bool {
return n/100 == 2
}
func Code2(n int) bool {
if http.StatusOK > n || n > 299 {
return false
}
return true
}
Go playground: https://go.dev/play/p/ARWCYyeR4rf.
Simply divide the HTTP code by 100 and compare it to an integer.
This can help you to avoid errors on the condition, missing an = or the < being on the wrong side etc.
Now that we have an excuse to compare performance let’s see how to write a Go benchmark.
Go already comes with a complete tooling to benchmark your functions.
code_test.go:
package main
import "testing"
func BenchmarkCode(b *testing.B) {
// run the Code function b.N times
for n := 0; n < b.N; n++ {
Code(n)
}
}
func BenchmarkCode2(b *testing.B) {
// run the Code2 function b.N times
for n := 0; n < b.N; n++ {
Code2(n)
}
}
Let’s look in detail at the signature of the functions:
BenchmarkCode(b *testing.B)
You can see that the input of the benchmark test is testing.B as opposed to a unit test which has testing.T.
To run your benchmark, use this command:
go test -bench=.
The loop on b.N allows us to set the number of iterations with parameters to the go test command using -benchtime=1000x.
You can also use -count 10 to run your test several times and get an average at the end.
Result of Code2 function
$ go test -benchmem -run=^$ -bench ^BenchmarkCode2$ test -count 10
goos: linux
goarch: amd64
pkg: test
cpu: Intel(R) Xeon(R) CPU E5-2689 v4 @ 3.10GHz
BenchmarkCode2-4 1000000000 0.3477 ns/op 0 B/op 0 allocs/op
BenchmarkCode2-4 1000000000 0.3361 ns/op 0 B/op 0 allocs/op
BenchmarkCode2-4 1000000000 0.3312 ns/op 0 B/op 0 allocs/op
BenchmarkCode2-4 1000000000 0.3594 ns/op 0 B/op 0 allocs/op
BenchmarkCode2-4 1000000000 0.3381 ns/op 0 B/op 0 allocs/op
BenchmarkCode2-4 1000000000 0.3759 ns/op 0 B/op 0 allocs/op
BenchmarkCode2-4 1000000000 0.3297 ns/op 0 B/op 0 allocs/op
BenchmarkCode2-4 1000000000 0.3923 ns/op 0 B/op 0 allocs/op
BenchmarkCode2-4 1000000000 0.3461 ns/op 0 B/op 0 allocs/op
BenchmarkCode2-4 1000000000 0.3335 ns/op 0 B/op 0 allocs/op
PASS
ok test 3.956s
Result of Code (the trick) function
$ go test -benchmem -run=^$ -bench ^BenchmarkCode$ test -count 10
goos: linux
goarch: amd64
pkg: test
cpu: Intel(R) Xeon(R) CPU E5-2689 v4 @ 3.10GHz
BenchmarkCode-4 1000000000 0.3229 ns/op 0 B/op 0 allocs/op
BenchmarkCode-4 1000000000 0.3264 ns/op 0 B/op 0 allocs/op
BenchmarkCode-4 1000000000 0.3440 ns/op 0 B/op 0 allocs/op
BenchmarkCode-4 1000000000 0.3449 ns/op 0 B/op 0 allocs/op
BenchmarkCode-4 1000000000 0.3391 ns/op 0 B/op 0 allocs/op
BenchmarkCode-4 1000000000 0.3276 ns/op 0 B/op 0 allocs/op
BenchmarkCode-4 1000000000 0.3180 ns/op 0 B/op 0 allocs/op
BenchmarkCode-4 1000000000 0.3219 ns/op 0 B/op 0 allocs/op
BenchmarkCode-4 1000000000 0.3540 ns/op 0 B/op 0 allocs/op
BenchmarkCode-4 1000000000 0.3352 ns/op 0 B/op 0 allocs/op
PASS
ok test 3.749s
Conclusion
After several runs of the benchmark we can see that the Code solution performs slightly better.