Stanford opens new online courses about Compilers and Automata

A few months ago Stanford University started a great initiative: it offered free online courses for everyone. Courses were mostly about computer science stuff (Natural Language Processing, Machine Learning, Computer Vision etc.). If you didn’t take part in these courses, then you must know that they are just like normal classes at the university. There are lectures (about 2 hours per week), tests after each week, final exams and rather demanding programming assignments. I enrolled to Natural Language Processing, but I have to admit that I couldn’t keep the pace and stopped following the course after about 3 weeks. Perhaps I wasn’t motivated enough, since I NLP has no connection to my research and I wouldn’t be able to use the gained knowledge in practice. Yesterday however my motivation jumped to maximum, because I got a newsletter about a whole bunch of new courses. Among them are two courses, which are of high interest two me: Compilers and Automata.

Recently I got very interested in the compiler construction subject. I got myself the Dragon Book, but it’s really huge so I think I wouldn’t read all of it. I also visited #compilers IRC channel and got some very valuable hints on how to dig into the subject of constructing a compiler. I was suggested to begin with reading some sort of “Lisp interpreter in Lisp” tutorial. I chose Scheme as my Lisp dialect and – knowing nothing about Scheme – I read the R5RS language specification (only 50 pages!) and The Little Schemer (chapter 10 shows how to do a Scheme interpreter in Scheme). I’ve also found Write Yourself a Scheme in 48 Hours tutorial, which shows how to write Scheme interpreter in Haskell. This is a bit more advanced and, despite what author claims, is not suitable for Haskell newbies like myself. I decided to get back to that tutorial after learning some more Haskell, especially monads and Parsec, both used heavily by the author. Guys at #compilers further suggested that after reading this kind of tutorial I should read Niklaus Wirth’s “Compiler Construction”. Indeed, the book looks very accessible. Another usefull advice I received was “the older the better” since fundamentals of compilers haven’t changed much and the only substantial progress was achieved in the back-end1 development. However, the most motivating was the estimation that writing a simple compiler from scratch should take about 6 months. Nevertheless, learning everything by myself from scratch was a very slow process and I wish I had someone who would teach me. The course on Compilers from Stanford is therefore a real blessing. Well, looks like my wish was granted.

The course about Automata is also something very exciting. I have a book “Introduction to Automata Theory, Languages, and Computation” by John E. Hopcroft, Rajeev Motwani and Jeffrey D. Ullman. I approached this book twice but always got stuck around third chapter. Not that I couldn’t understand it. It was my motivation that failed. Anyway, I feel that as a computer scientist I should have a basic knowledge of this subject. Believe it or not, but my studies didn’t cover the subject of automata and computations!

There’s a lot of other courses, not only about computer science, so I suggest to take a look. Perhaps you’ll find something interesting for yourself. Participating in NLP course had showed me that these courses are very demanding and require a lot of time and effort. This means that I’m going to pick only the Compilers course and concentrate solely on it, hoping that the Automata course will be repeated in the future. Both courses start this Monday (April 23rd).

  1. back-end is the part of compiler responsible for generating the final code (most often a machine code) []

One Response to “Stanford opens new online courses about Compilers and Automata”

  1. Hugo Estrada says:

    Thanks so much for sharing the link to the class. I went ahead and enrolled in the compiler class.

Leave a Reply

(required)

Staypressed theme by Themocracy