Concepts, Techniques, and Models of Computer Programming

Ever wondered why it takes so long to pick up your first programming language, when it's C++ or Java? Ever wondered why Object-oriented Programming feels so difficult to get right, even after years of experience? This books is an amazing study of various programming paradigms (or models, as the authors call them). It starts with the most minimal features required in a programming language, discusses their impact on how you write small programs and then moves on to bigger concepts. Until you've read this book, you might not realise that multi-threaded object-oriented programming is such a powerful model that it can be used to easily write a lot of real-world applications but this power also makes it tough to master the model because of the many ways you can abuse it. The more powerful a model gets, the more difficult it becomes to verify its correctness without additional tools like debuggers, profilers, etc. Most importantly, this book can teach you two important things: * Multi-paradigm programming is more natural (i.e. easier to understand and model real-world concepts in) than 'pure' programming * Use the least powerful model that can solve the problem at hand naturally (i.e. you don't end up writing a lot of code to work around the model's limitations) A third thing that they don't enumerate but imply quite obviously is a program design methodology that involves writing large parts of the application using a less powerful and more deterministic model, while harnessing the power of more capable models only for those few components of the application that absolutely need them. The popular "shared-nothing" architecture for web applications, backed by a concurrent shared-state store (RDBMS, mostly) is one example of such an approach. The only shortcomings of this book that I found were the rather difficult installation of Mozart programming environment used to illustrate the book's concepts, and IMHO a shortage of sample problems that illustrated the usage of more advanced models.

This book is fantastic! It's like a more "fleshed out" version of Structure and Interpretation of Computer Programs - 2nd Edition (MIT Electrical Engineering and Computer Science). It uses a neat (if somewhat weird) language called Oz, which has a number of interesting features, which are used to demonstrate the concepts of the book. Much like SICP, this book is a real masterpiece, elegantly composed and explained.

Modern programming has become fragmented into a variety of computational models (OO, functional, imperative, etc), and a variety of languages supporting those computational models. Neophyte programmers are typically introduced to just one of these models, and only learn the other, "less natural" models later. With CTM, Van Roy and Haridi take an alternative approach. They teach programming as an integrated discipline, and demonstrate the underlying links between the different computational models. By the time the reader is done with the book they will have a much better understanding of the discipline of programming, and will be well-equipped to decide which model is best suited to the task at hand. Reading CTM is an extremely worthwhile experience for anyone wishing to achieve a deep understanding of the art and science of programming. CTM has been compared to Abelson and Sussman's "Structure and Interpretation of Computer Programs". They are similar, in the sense that they both provide the reader with a deeper understanding of programming than most programming texts. However, the content of both books is quite different, and it is definitely worth reading both. Another book that I feel makes a good companion to CTM is Hoare's (sadly hard to come by) "Unifying Theories of Programming". It covers a lot of the same material as CTM, but in a much more theoretical sense. Where CTM is concerned with practical programming, Hoare is concerned with mathematical underpinnings. The two complement each other nicely.

This book is a real mind-bender that illuminates paths for computer design at both the conceptual and practical levels I'd never travelled down before.The notion that one language can be so flexible as to accomodate both the syntax and semantics of so many different computational models, or paradigms, took some unlearning of bad programming practice before its power, elegance and potential began to sink in.It also explodes the myth that "pure" languages -- i.e., pure OO, or pure functional, etc., languages--have some kind of innate advantage over so-called "hybrid" languages. In fact, "hybrid" (or as the authors would prefer to call them, "multi-paradigm") languages come out of this book looking even more powerful than the "pure" ones, insofar as they allow the programmer to use the right model for each task, instead of trying to make OO fit, for instance, in places where it doesn't fit so well. The idea here is that each computational model represents a completely different way of approaching a domain problem. Used by themselves, each has its niche. For instance, everybody knows OO is good for domain modelling and busines objects. Prolog-type languages are good for applications that need to apply rules over a set of data. Functional languages are great in mathematical applications. And so on. What is new here is that one can program in an environment in which all of these tools are available in a single core semantics that seamlessly weaves these computational models into a complementary whole. Used together judiciously, with an eye toward program correctness, they make things possible that have long been considered very hard -- for instance, constraint programming. Mozart-Oz, the underlying technology, is a strange language when you first look at it. It's hard at first to get used to concepts like "higher-order programming" or "by need execution" or "lazy execution" if you are the programming grunt in the field of most modern IT shops, forced by bosses to code in your standard fare -- Java, C#, VB, etc. If OO in Java is like the hammer that makes everything look like a nail, in Mozart-Oz you have a language that is like walking into Ace hardware store, a swiss army knife of a language (conceptually speaking) that challenges you to become a highly skill code craftsman, not just a programmer.But, if only for the personal growth you will experience grappling with the concepts in this book, I recommend it very highly even to "non academic" programmers (like myself) as well as to any advanced student of computer science. It may be painful, you may scratch your head in places where the concepts just seemed to leap over your cranium, but if you are patient, do the exercises (and at least think about what it would take to tackle some of the research projects), you will grow.Unfortunately, you may find the languages you work on to be rather confining, and maybe even boring, after you get a whiff of what multi-paradigm programming can do. More likely, ho

In 1976 Edsger W. Dijkstra elevated programming to an intellectual discipline and taught us how to reason about what we now call "imperative programming". To illustrate his methodology Dijkstra solved challenging problems with unforgetably beautiful, yet simple and powerful example programs that are as relevant today as they were forty years ago. Since then, programming has splintered into paradigms, methodologies and suffers from baroqueness, perpetuation of obsolete conventions and other practices that restrict the full expressive power of programming "as a whole".In 2004 Van Roy and Seif Haridi have given us a glimpse of what programming can be like without unnecessary restrictions imposed by paradigms and other heavy baggage caused by politics, ideology and historical inertia. Using the remarkably mature implementation of the Mozart system and the conceptually clean, simple, elegant, yet powerful programming language Oz, Van Roy and Haridi show us how dogmatic heavy baggage falls away when we can look at programming as a whole and choose the best programming concepts that the solution of a problem requires. Such a program becomes simpler, more elegant and therefore less error prone than an equivalent solution that is restricted to a specific paradigm.