Haskell, improved

I showed my Haskell program to a co-worker today. As we were looking at it, we noticed some possible improvements. (Surprising, since neither of us are Haskell programmers.)

I tried the changes this evening, and they do indeed work. We've simplified the program and made the logic clearer. I also expanded the program to handle a list of payment values, and not assume that each payment is the same value. It was quite easy to do -- much easier than similar changes to a C++ or C# program.

Haskell for real

Today was the day that I wrote a 'real' -- that is a non-trivial, example-from-the-book -- program. It was an exhilarating experience.

The program (actually three functions) computes the price of a stream of payments.

I was inspired by a task at the office: review code that computes the price of the stream of payments. The code was in two forms: Java and Fortran. The code also came with the math for the calculations, and I immediately thought of functional programming.

Today I set aside some time and dug out the one book on Haskell in my library. It had enough to get me started. With the book and the math, I was on a programming adventure!

Programming in Haskell is a trip. Just as object-oriented programming was alien to my brain trained in procedural programming, functional programming is alien to my brain now trained in object-oriented programming. I struggled a bit with the syntax and with the design of the program, and emerged victorious! I had a set of functions that performed the desired calculations. (I checked them with math calculated in a spreadsheet.)

The Haskell version is shorter and cleaner than the procedural versions. (The Fortran version is obviously procedural, and the Java version makes no use of object-oriented constructs.) The procedural version uses a loop and a temporary variable that is assigned at the end of the loop so that it can be used the *next* time through the loop.

The Haskell version has no loops and no assign-at-the-end distraction. If you know the syntax, it is straightforward and easy to understand. I believe it to be less efficient than the procedural versions (efficient meaning 'less CPU time') so with Haskell I have gained programmer time at the expense of CPU time, a trade-off that I am willing to accept.

I will say that I don't fully understand the Haskell version. That is, I am not fluent in the syntax. I have internalized the syntax of Fortran, C, Java, and other languages so I can look at the code and know what it does without thinking. I am still in the "think about the code" stage with Haskell. (And that's OK. I expect a ramp-up time with a new language.)