Saturday, September 02, 2006

Functional Programming Considered Harmful? Nonsense!

Luke Plant recently wrote a blog entry on "Why learning Haskell/Python makes you a worse programmer". He rants about how difficult it is to write in a functional way in C# (and Java as well). And ends with this quote:
The moral of the story: don't bother improving yourself, unless you have the freedom to improve your environment accordingly. That's rather depressing. Can anyone put a better spin on this and cheer me up?

After reading this article, I was reminded of an argument to remove objects and blocks from Smalltalk. Basically, the argument of closures versus objects is futile. They both have their place and strengths in design. But, without closures, you can use simple objects as a substitute.

Luke's article argues how hard it is to instill functional programminging in languages like C# and Java. I think while it is hard to use pure functional style in these languages, it's matter of changing the way you think about it. I find since I don't have blocks/closures in Java, I use more of a stream approach (think pipe/filter pattern). It's a little trickier in the iterator objects, but the users of the code have to write less and their code is easier. I used this technique to good effect in my Reflective Testing Framework. As for side-effect free programming, I think it's a good goal to try to attain. We have a thing called encapsulation which can localize the effects of change. It's all in the way we think about it.

So, the next time, you see a road block because Java or C# doesn't have some feature from one of your favorite dynamic or functional languages. Think about the goal of the feature and ways you can attain it. Sure, code is more succint in Ruby, Python, Haskell, Scheme, and a host of other languages, but for those of us in the trenches, we might not be able to use them. We're stuck with Java/C#/C++. Be creative and remember to always keep your code easy to understand.

1 comment:

David Farber said...

Heh heh. I don't think anyone ever understood what I wrote--though that's probably nobody's fault but my own--except that it didn't help that Vassili immediately spoofed my message. The point I wanted to make was that blocks and methods have overlapping functionality, and that Smalltalk could conceivably be simpler (and thus be per capita more powerful) if you removed the overlap.

So, my post wasn't really about removing Blocks from Smalltalk (despite the title--which is the only thing most people read).

Amazon