natural

Yet another natural sort, with 100% test coverage and a benchmark. It does not allocate memory, doesn't depend on package sort and hence doesn't depend on reflect. It is optimized for speed.

Benchmarks

On Go 1.18.3.

$ go test -bench=. -cpu 1
goos: linux
goarch: amd64
pkg: github.com/maruel/natural
cpu: Intel(R) Core(TM) i7-10700 CPU @ 2.90GHz
BenchmarkLessDigitsTwoGroupsNative 331287298     3.597 ns/op   0 B/op   0 allocs/op
BenchmarkLessDigitsTwoGroups        32479050    36.55 ns/op    0 B/op   0 allocs/op
BenchmarkLessStringOnly            157775884     7.603 ns/op   0 B/op   0 allocs/op
BenchmarkLessDigitsOnly             69210796    17.52 ns/op    0 B/op   0 allocs/op
BenchmarkLess10Blocks                6331066   190.8 ns/op     0 B/op   0 allocs/op

On a Raspberry Pi 3:

$ go test -bench=. -cpu 1
goos: linux
goarch: arm
pkg: github.com/maruel/natural
BenchmarkLessDigitsTwoGroupsNative  14181789    86.57 ns/op    0 B/op   0 allocs/op
BenchmarkLessDigitsTwoGroups         1600195   748.9 ns/op     0 B/op   0 allocs/op
BenchmarkLessStringOnly              8286034   142.3 ns/op     0 B/op   0 allocs/op
BenchmarkLessDigitsOnly              3653055   331.4 ns/op     0 B/op   0 allocs/op
BenchmarkLess10Blocks                 310687  3838 ns/op       0 B/op   0 allocs/op

Coverage:

$ go test -cover
PASS
coverage: 100.0% of statements
ok     github.com/maruel/natural       0.012s