Refactor trailer Field for Improved Memory Efficiency and Performance

[ksw2000]

The original trailer field was represented by []argsKV, where argsKV consisted of two []byte slices and one bool, totaling 56 bytes. However, maintaining each element required only a single 24-byte slice.

After refactoring, there was a significant reduction in memory usage.

I designed two benchmarks to compare the performance between the refactored code and the original implementation.

• BenchmarkSetTrailerBytes primarily measures the performance of the AddTrailerBytes function.
• BenchmarkSetTrailerBytesWithReset evaluates the efficiency of reusing memory by repeatedly adding trailers and resetting.

func BenchmarkSetTrailerBytes(b *testing.B) {
	list := make([][]byte, 5)
	for i := range list {
		list[i] = []byte(fmt.Sprintf("trailer-%d", i))
	}
	for i := 0; i < b.N; i++ {
		header := RequestHeader{}
		for j := range list {
			header.AddTrailerBytes(list[j])
		}
	}
}

func BenchmarkSetTrailerBytesWithReset(b *testing.B) {
	list := make([][]byte, 5)
	for i := range list {
		list[i] = []byte(fmt.Sprintf("trailer-%d", i))
	}
	header := RequestHeader{}
	for i := 0; i < b.N; i++ {
		for j := range list {
			header.AddTrailerBytes(list[j])
		}
		header.Reset()
	}
}

goos: linux
goarch: amd64
pkg: github.com/valyala/fasthttp
cpu: Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz
                           │   old.txt   │           new.txt           │
                           │   sec/op    │   sec/op     vs base        │
SetTrailerBytes-8            966.6n ± 1%   756.6n ± 1%  -21.72% (n=50)
SetTrailerBytesWithReset-8   202.2n ± 0%   176.0n ± 1%  -12.93% (n=50)
geomean                      442.1n        365.0n       -17.44%

                           │   old.txt    │               new.txt                │
                           │     B/op     │    B/op     vs base                  │
SetTrailerBytes-8            944.0 ± 0%     456.0 ± 0%  -51.69% (n=50)
SetTrailerBytesWithReset-8   0.000 ± 0%     0.000 ± 0%        ~ (p=1.000 n=50) ¹
geomean                                 ²               -30.50%                ²
¹ all samples are equal
² summaries must be >0 to compute geomean

                           │   old.txt    │               new.txt               │
                           │  allocs/op   │ allocs/op   vs base                 │
SetTrailerBytes-8            10.00 ± 0%     10.00 ± 0%       ~ (p=1.000 n=50) ¹
SetTrailerBytesWithReset-8   0.000 ± 0%     0.000 ± 0%       ~ (p=1.000 n=50) ¹
geomean                                 ²               +0.00%                ²
¹ all samples are equal
² summaries must be >0 to compute geomean

The refactoring of the trailer field from []argsKV to a single slice significantly reduced memory usage up to 51.69% and improved execution time up to 21.72%.