Saturday, June 18, 2011

SICP -- Starting with the P2PU study group

We have finally started with the SICP study group and I'm delighted to put some time into it. For once, I will focus all my effort and free time to work through and enjoy this fantastic book.

The goal of the group is to finish one section every second week, which gives us 6 weeks for the first chapter. Some people might think that 6 weeks is a lot but to be honest, at least for me, with a full time job, a part time PhD to work on, and all the normal social commitments, it's not that much time. So let's get cracking!

I decided to go back a re-watch the first bunch of MIT SICP lectures as it's been a while since I got around to see them first. I always have this funny and a bit weird feeling when I start watching one because for a moment I feel like I am watching an 80's sitcom or something similar. I'm probably just being silly but it always makes me smile!

The lecture series starts with Harold Abelson ranting a bit about the term Computer Science not being a very good one when it's not a real science, or even about computers. I do agree with him but if you want to know more about the subject you will have to watch it for yourself.

More interesting are his comments about the distinction between declarative or descriptive knowledge and imperative or procedural knowledge. Procedural knowledge or 'Know-How' can be directly applied to a task, whilst declarative knowledge tells us about truth. His example for this is the fact that the square root of a number can be expressed in the form of a mathematic equation(descriptive), compared to a way or algorithm to finding that square root through approximation (imperative).

He also explains that although you can learn most of the LISP language (its rules anyway!) in a short lecture, that does not make you a good user of the language. He compares it to learning the rules of chess and being a good chess player; you need the rules to play, but the rules do not make you any good at it. So you need to embed yourself in the context and the experience, and the techniques to do so are the topic of the course.

In Abelson's words, programming is all about managing complexity, and the techniques in this course will allow you to do so.

The lectures are divided into three main topics which do not correspond to the five sections of my book (my edition is from 1996), but that can be easily correlated, or so I think.

The parts discussed in the lectures are:
    - Black-box Abstraction: I guess this will be chapters 1 and 2, building abstractions with procedures/data.
   - Conventional Interfaces: I'd say this is chapter 3: Modularity, Objects, and State.
   - Metalinguistic Abstractions: This is chapter 4, of the same title.

The fifth chapter of the book about Register Machines does not seem to be in the lectures, but I couldn't tell right now cause I have yet to watch them all.

And that is all for now. I will spend another couple of hours later this evening working through the book and preparing the meeting for tomorrow Sunday at 7pm GMT, 3pm EST. Meetings happen on the #sicp channel on freenode IRC and everybody is welcome!!!

No comments:

Post a Comment