Reason for offloading file hashing to git?

[markjm]

Hi - looking at GitHashForFiles, I wonder why we want to use this method of spawning a process and letting git calculate the hashes for these files. Perhaps git stores some of this information internally so it doesn't need to be recalculated? Still not sure the overhead is worth it.

For example, in my test project with a large globalDependencies set (which may be a common transitory scenario for a large monorepo as it transitions to be more cacheable), I have benchmarked:

GitHashForFiles ~7seconds

But replacing it with the git-like hashing you have already built

func GitHashForFiles2(filesToHash []string, PackagePath string) (map[string]string, error) {
	changes := make(map[string]string)
	for _, file := range filesToHash {
		hash, err := GitLikeHashFile(file)
		if err != nil {
			return nil, fmt.Errorf("hash failed: %w", err)
		}
		changes[file] = hash
	}
	return changes, nil
}

GitHashForFiles2 ~.6seconds