How to Design Classes (in Ruby, Python, Java or any OOP) | Hacker News Reader

How to Design Classes (in Ruby, Python, Java or any OOP)

ccs.neu.edu

112 pointsOpen-Juicer24 comments

This looks like a great book, but I'm not sure it really represents a good way to learn how to design classes in Ruby or Python. It seems heavily biased towards designing classes in a statically-typed, non-type-inferring, java-style language. Programming in Python is a lot more about knowing when defining a new class (rather than using built-in types and classes) even a good idea to begin with.

I shamelessly clicked away as soon as I realized it was Java. Well, not completely shamelessly. I kind of feel like a "rassa-frackin' whippersnapper" disgustedly changing the radio station when stumbling on classical orchestral music.

Then again I also feel a little dirty for comparing the likes of Tchaikovsky and Beethoven to Java...

This book takes a theoretical approach to OOP.

As the author said, this book is NOT about Java. Similarly, SICP is NOT about Scheme.

Interestingly, It's the same author who teaches how to design programs with Racket, a lisp dialect.

http://www.ccs.neu.edu/home/matthias/HtDP2e/index.html

The book opens with "What you'll learn: Java."

You'll learn Scheme going through SICP too.

From the author:

"It is a good idea to study the programming language that you use on a daily basis and to learn as much as possible about it. We strongly believe, however, that it is a bad idea to teach the details of any programming language in a course. Nobody can predict which programming language you will use.

Therefore, time in a course is better spent on studying the general principles of program design rather than the arcane principles of any given programming language."

It actually says "What you won't learn: Java."

In my defense, this is what shows up on my screen:

  What you will learn
  What you won't learn
  Java.

Weird!

Your defense is that you badly misquoted what shows up on your screen?

That is weird...

Honest question: Is that really what you thought? It seems obvious to me that I was implying a defense that the phrase I quoted also appears on the page, directly next to the phrase ekiru quoted, which makes the mixup a little more understandable, because it means I didn't see something that simply didn't exist. Did you get that and post your reply anyway, or did I just express myself poorly?

> As the author said, this book is NOT about Java.

That's what it says, but the concepts covered are so java-oriented it may as well be.

It uses generic language to describe most of the concepts, but the underlying assumption seems to be that you need to understand these concepts because you need to program in Java, C#, C++, or some other closely-related language with no dynamic typing, no first-class functions, and explicit java-style semantics for abstract classes, interfaces, public vs private methods, implements vs extends, and such.

Much of that is invisible in a language like Python, and a book that laboriously covers each concept in the context of a language where you have to specify everything explicitly probably isn't going to make you much better at writing classes in python. Or if it is, it's going to be a lot of work for minimal benefit.

> Interestingly, It's the same author who teaches how to design programs with Racket, a lisp dialect.

Indeed, that's partly why I'm surprised the book takes the approach it does. However, it does seem to be a good, thorough book (ignoring the missing sections) if you need to design classes in Java or C# (which covers a lot of people).

I think this is where a lot of Rubyists/Pythonistas get lost. Just because the languages hide ideas like "interface" and "abstract base class" doesn't mean that those concepts aren't useful. Particularly in Ruby-land there's been a push to have common interfaces. They're only enforced by contract. And if you think about it, having "interface" explicitly in the language just means that the contract is enforced at compile time. Even so, you have to be careful. You could implement an interface and still just return null.

> I think this is where a lot of Rubyists/Pythonistas get lost. Just because the languages hide ideas like "interface" and "abstract base class" doesn't mean that those concepts aren't useful.

I agree, somewhat. The problem is that HtDC seems to spends so much time on vocabulary and describing how to construct classes that very little is left over for discussion of actual design. Here's an excerpt:

This first section on classes suggests that the design of a class proceeds in three steps:

1. Read the problem statement. Look for statements that mention or list the attributes of the objects in your problem space. This tells you how many fields you need for the class definition and what information they represent.

That is probably quite reasonable from a java-perspective. But in Python the best answer might be to use the built-in data structures, define a bunch of functions to operate on that data, which are automatically encapsulated in a namespace based on the name of the file. But it might not, sometimes you would want to define a class with explicit fields.

The question is, how do you know which technique to use? This book doesn't seem to help with that. We've jumped from a data-driven approach using scheme to a java-style object-oriented approach for no apparent reason and very little explanation of the benefits of this approach.

Even if the part about interfaces might help a bit with designing a Python module, in order to get that out of this book you're going to have to wade through a lot of other stuff.

No, this is quite right. Classical music is as bloated, heavy-handed, and decadent as Java. Good music is found in the 16th century and earlier, and in the 20th century and later. We can see this maps to software: good ideas are to be found in the 70s and earlier, and in some hoped-for new era yet to come . . .

You need to go get yourself a recording of Rubinstein playing Chopin's Nocturnes. Hurry!

EDIT: I seek no arguments. Wikipedia says nocturnes are generally thought of as "expressive and lyrical". Surely that should be of interest to hackers...

Check out Haydn's "Lord Nelson" mass, specifically the recording by Banchetto Musicale (it's very light and clear). For the antithesis of heavy-handedness, pay attention to movements 2, 4, 5, 7, and 12. But the whole thing is only about 40 minutes and really easy enjoy start to finish, especially since those movements I mentioned are meant to contrast with the slower ones.

Movement 5, while not my favorite, is especially notable for being a canon at the 5th. It's quick, fun, the imitative counterpoint is very easy to hear, and at a minute-forty, hardly overstays its welcome.

Overall though, Western music between Bach's death in 1750 and Beethoven was anything BUT bloated and heavy-handed. Balance, taste, and clarity were highly valued.

This book is rather naive. It starts with a quote by, the inventor of Smalltalk, Alan Kay. The book focuses on Java.

Smalltalk is a message-passing language. Java is a method-calling language.

A book about object-oriented programming that doesn't distinguish between message-passing and method-calling fails to express how languages like Ruby and Objective-C work.

Excellent point. I think that more generally, as it relates to what a student might be interested in, it fails to address the fundamental question "why am I learning this?"

I think if I were going to design a book or course in object-oriented programming I would either approach top-down using a dynamic language or bottom-up with C. In the top-down version I would focus on the advantages of OOP as a discipline and why seemingly arbitrary restrictions really are useful (not just some hand-waving about static type checking) Going the other way, I'd have them write method dispatchers and basically implement classes without the features of modern object-oriented languages and then discuss OO languages in terms of the features they provide to simplify that style.

This book is too long for the subject it purports to impart.

Good coding is not a craft of many secrets; rather it is an art of a few core principles. Mastering them is a matter of direct experience. I believe a good book can point a programmer in the right general direction, but it is up to her to go the distance.

Compared to bicycles and shapes, this book is a step in the right^H^H^H a different direction in that it is data-oriented rather than object-oriented. Rather than make the claim (as they do in the introduction) that other textbooks are doing it wrong, I would have prefered to see acknowledgment that different tasks call for different approaches, and highlight that the data-oriented design presented here is usually left out of other texts. Giving examples, as they do, of tasks where the goal is simply to store and present data, it is only natural that a data-oriented design is what shows up. That does not mean that there are not programs whose job is to act on and manipulate data.

But. Fundamentally, designing classes upfront is simply the wrong way to go about it. The argument may be made that the end result (I now know what good classes look like) justifies the method, but I disagree. It is the journey that is important, not the result. The students will not remember what a good jogging journal class hierarchy is. They will remember how to design classes on paper. In my experience, classes hierarchies written on paper or on fancy UML software, are bound to be "wrong". This is a Bad Thing.

UPDATE: In fact, any tutorial about how to design classes that does not describe an iterative process, and make use of all the refactorings built into Eclipse or IntelliJ, is guaranteed to fail to teach kids what I do every day I program java (or C#).

This is the textbook for one of my classes at Northeastern. The prerequisite class used HtDP (http://www.ccs.neu.edu/home/matthias/HtDP2e/) and Racket (http://racket-lang.org/), and the focus was on design principles and writing good code more than it was on the actual language we were using. I'm expecting the same from this class, but I'm not against becoming more experienced with Java either.

I found this to be a great (ruby based) design book. Ruby best practices by Gregory Brown (free download http://rubybestpractices.com/)

Eloquent Ruby is the design book to read.

Why not import some of HtDC best ideas to the Ruby world, for those of you think HtDC is Java biased? Actually, many good ruby books are adapted from the Java land.

A side note: To many rubyists' surprise, Matz is not a language-biased guy who would like to start a language war.

666 pages.

Nice to see NEU on here, they are big Scheme fans there I know so I don't know how much Ruby it will relate to. Python is used in some classes as well I believe (I was an ME major with friends in CS).

Reply on news.ycombinator.com