This is a mobile version, full one is here.

Yegor Bugayenko
11 April 2017

Flexibility Equates to Lower Quality

There are two opposing mindsets: "If it works, it's good" vs. "If it's good, it works;" or "Make it work" vs. "Make it right." I'm talking about the software source code. I've been hearing this almost every day in blog comments: Why do we need all those new OOP principles if our code works just fine without them? What is the point of introducing a new way, which is supposed to be "better," if the existing, traditional, semi-object, semi-procedural, not-so-perfect approach just works?

Let's try to think bigger. And not just about object-oriented programming, but in general about software development. There are many examples of the "just works" mentality.

Take Perl, a programming language famous for its ability to do anything in three different ways. This means that there is no one "right" way. I'm not a Perl expert; that's why I'll have you look at this Ruby code instead:

if a > b
  m = 'Hello!'
end

We can rewrite it like this:

m = if a > b
  'Hello!'
end

Or this:

m = 'Hello!' if a > b

And one more:

m = a > b ? 'Hello' : nil

Which one is "right?" Are there any Perl developers? Can you suggest some other way to achieve the same result?

Not surprisingly, in Java (a stricter language than Ruby), there is only one way to do it:

if (a > b) {
  m = "Hello!";
}

Well, I guess I'm wrong; there are two, actually. Here is the second one:

if (a > b) m = "Hello!";

What does this variety of options give us, as programmers? I guess the answer seriously depends on what we, the programmers, are doing with the code: writing or reading it. Also, it depends on our attitude toward the software we're creating; we either own it (hacker mentality) or build it (designer mentality).

If we're writing it, and we love to think about ourselves as code owners, we definitely will need that arsenal of syntactic sugar weapons. We need them to prove to ourselves that we're smart and, of course, to show off in front of our friends and that soulless Ruby interpreter.

On the other hand, if we're designers and happen to read the code that is full of sugar, which "just works," we'll be very annoyed and frustrated. Well, maybe I have to speak for myself, but I definitely will be.

This overly-sugared Ruby syntax is a perfect example of "works vs. good" positioning. Ruby philosophy is this: It doesn't matter how you write it, as long as it works. Java philosophy is different; it's much closer to: Make it right and it will work. The weak and dynamic typing in Ruby vs. the strong and static one in Java also prove my point.

In general, I believe that the more flexible the programming language is, the lower the maintainability—the key quality characteristic—of the code it produces. Simply put, higher quality comes from simpler languages.

The same is true for the entire software development: The more restrictions we put on programmers and the fewer options they have for their "syntax creativity," the higher the quality of the software they write. Static analyzers like Checkstyle for Java or Rubocop for Ruby attempt to solve that problem and prohibit us from using certain language features, but they lag far behind. We are very "creative."

Now, let's get back to the original OOP question: Why do we need to improve anything if it works the way it is? Here is the answer: Modern OOP (as in Java, Ruby, and C++) doesn't produce quality code because it doesn't have a strong and properly restricted paradigm. It just has too many "features," which were mostly introduced by C++ and remained there for our mutual "convenience."

They indeed work, but the maintainability of the software we produce is very low. Well, it's way lower than it could be, if our "creativity" would be restricted.