plumbing: object, Optimize logging with file.

[onee-only]

Connected to: #811

So, I've been digging this issue. Then I found that object.(*treeNoder).NumChildren was the one that is consuming most of cpu time.

After some investigtion, I figured out that treeNoder.children field is documented as "memoized", but none of the code actually saves its result.

type treeNoder struct {
	parent   *Tree  // the root node is its own parent
	name     string // empty string for the root node
	mode     filemode.FileMode
	hash     plumbing.Hash
	children []noder.Noder // memoized <- here!!
}

So I modified the code to save the result of transformChildren(). And here's how it is changed.

Ran on Apple M2 processor

Before

Output:

$ go run main.go
2024/03/10 13:53:55 cloning repository
2024/03/10 13:54:01 git log start
2024/03/10 13:54:33 time elapsed: 31.432235916s

After

Output:

$ go run main.go
2024/03/10 13:49:51 cloning repository
2024/03/10 13:49:58 git log start
2024/03/10 13:50:10 time elapsed: 12.343892125s

I'm guessing there could be more optimization like making object.(*Tree).Decode faster, but I have no idea what to do with it.

Also, I added little optimization in (*commitPathIter).getNextFileCommit() which reuses parent's tree in iteration.

code used to test it:

package main

import (
	"log"
	"time"

	"net/http"
	_ "net/http/pprof"

	"github.com/go-git/go-billy/v5/memfs"
	git "github.com/go-git/go-git/v5"
	"github.com/go-git/go-git/v5/plumbing/object"
	"github.com/go-git/go-git/v5/storage/memory"
)

func main() {
	go func() {
		log.Println(http.ListenAndServe("localhost:6060", nil))
	}()

	fs := memfs.New()

	log.Println("cloning repository")
	rep, _ := git.Clone(memory.NewStorage(), fs, &git.CloneOptions{
		URL: "https://www.github.com/sigmahq/sigma",
	})

	f := "README.md"

	log.Println("git log start")
	start := time.Now()

	cIter, _ := rep.Log(&git.LogOptions{FileName: &f})
	_ = cIter.ForEach(func(c *object.Commit) error { return nil })

	log.Printf("time elapsed: %s\n", time.Since(start))
}

[pjbgf]

Tested these changes by modifying the log example, the time to run dropped from ~18s to ~6s:

Before:

$ go run _examples/log/main.go
git clone https://github.com/src-d/go-siva
git log
git log start
2024/03/10 07:17:00 time elapsed: 18.761931ms

After:

$ go run _examples/log/main.go
git clone https://github.com/src-d/go-siva
git log
git log start
2024/03/10 07:17:28 time elapsed: 5.999793ms

[pjbgf]

LGTM

Thanks @onee-only! 🙇