Concepts, Techniques, and Models of Computer Programming

Concepts, Techniques, and Models of Computer Programming, otherwise called CTM, by Peter van Roy and Seif Haridi is my "one book" for software development, if you need to pick one.

What I love about CTM is that it covers a huge range in topics, from programming in the small to in the large, functional programming to object oriented programming, and sequential programming to concurrent programming. Not only does it cover all these topics but, in my opinion, covers each of them sufficiently such that the reader can understand them and successfully use them correctly.

I don’t remember exactly when I discovered CTM but probably three to five years after it was published (which was in 2004). It’s 900 pages. Definitely a text book. I have a memory of somewhere between 2007 and 2009 reading about how ports (channel for messages) were implemented. I struggled but after I understood it I was both very proud and felt I’d learned something new, and different, and something that I hadn’t seen in mainstream languages before and still haven’t.

Where I remember CTM, and Mozart/Oz, really shining is in distributed systems. Distributed systems are especially difficult when it comes to failures and CTM made failures tractable. Its usage of failure streams, in particular, is a concept I have not seen anywhere, despite it being very powerful and useful. Mozart/Oz isn’t stuck in the traditional RPC world that seems so good at first but breaks down in the face of latency and failures. It provides a model that lets you get a transparent-feeling RPC interface but also gracefully handles errors. Its approach to distributed programming goes far beyond that, all while staying correct and simple (or as simple as possible).

While I think CTM is a profound piece of work, it does not cover all elements of software development that I think are important. I am very much a proponent of statically typed languages, such as Ocaml. Mozart/Oz is a dynamically typed language. But that’s OK, there is only so much room. I think that types are a valuable part of software engineering but the concepts CTM covers are independent of types and a side journey into types would likely obfuscate the discussion. The value of CTM is not necessarily using every concept in the book, but being aware of them, understanding them, and using them to inform your own software development.

My goal is to read, on average, two chapters per week and write a short post based on those chapters. I am not a skilled enough writer, and I do not understand the concepts of CTM well enough, for any of these blog posts to be a replacement for actually reading CTM, nor is that the goal, but I think this book is simply so great that it’s worth reading and thinking about deeply enough to write about.

Up next: Chapters 1 & 2.