Reading Code

I enjoy reading code and want to talk about it.

At my last job I was responsible for doing code review for nearly all the code that we released. The only code I didn’t review was the realtime C that was released (somewhat) out of band. This meant that I reviewed lots of Perl, lots of JavaScript, and good amount of C#.

I remember getting good enough at it that one time I caught a race condition in the C#, mentioned that it was a possiblity, but the team decided to release anyway and fix it later. Within days a customer experienced the race condition.

All that said, I have gained an appreciation for reading code. I am sure that intense code review isn’t the only way to gain a love of reading code, but it did the trick for me.

Now I will often be reading documentation and get curious and read how some code is implemented. I especially appreciate how Go’s documentation links directly to the implementing code, which makes diving into the code even easier.

Just yesterday I was reading some of spilld and came across this delightful trick for creating MIME bondaries:

func randBoundary(rnd *rand.Rand) string {
	var buf [12]byte
	_, err := io.ReadFull(rnd, buf[:])
	if err != nil {
		panic(err)
	}
	// '.' and '.' are valid boundary bytes but not valid base64 bytes,
	// so including them provides trivial separation from all base64
	// content, which is how all tricky content is encoded.
	return "." + base64.StdEncoding.EncodeToString(buf[:]) + "."
}

Here’s a super useful trick to log AWS API calls I ran across while reading some code at ZipRecruiter:

	sess, err := session.NewSession(cfg)
	if err != nil {
		return errors.Wrap(err, "creating a new AWS session")
	}

	if *debug {
		sess.Handlers.Send.PushFront(debugHandler)
	}

// ...

func debugHandler(r *request.Request) {
	fmt.Printf("Request: %s/%+v, Payload: %+v\n", r.ClientInfo.ServiceName, r.Operation, r.Params)
}

I don’t want this post to become some kind of gallery of clever code. I just find it surprisingly often that I will be casually perusing code, sometimes to see how something works because I want to use the same technique, other times because the documentation isn’t quite clear, and I’ll find myself enjoying the process. I think part of it is cunning tactics (like the MIME bit above) that allow removing huge swaths of code, but often it gives me an insight into other minds.

I think reading code is a skill all software engineers need to develop and improve. We spend so much of our time editing existing code that it’s important for us to orient ourselves and make the correct modifications. It’s not good to make changes that are contrary to the goals of the code we are updating, unless we need to change those goals, which probably implies much bigger refactors anyway.


If you read this blog you probably either are interested in, or write, Perl or Go. If you aren’t already well-versed in reading code, I would suggest practicing!

If you are new to Go, I would first read a primer on the language like The Go Programming Language or maybe Go Programming Blueprints. After that I would dive into the wealth of examples that is the standard library! A great starting point (some of the first Go that I read and really appreciated) would be the io.Copy function; click the function name to get started and keep reading till you understand it.

I suspect the Perl prorgrammers new to this blog are few and far between, but if you are, check out Programming Perl. I haven’t read Perl code in quite a while, but some good starting points might be Plack, Moose, and DBI.

Posted Tue, Mar 26, 2019

If you're interested in being notified when new posts are published, you can subscribe here; you'll get an email once a week at the most.