Core Java

Are You Still Debugging?

Debugging is “a process of running a program/method interactively, breaking execution flow after each statement and showing…” In a nutshell, it is a very useful technique … for a bad programmer. Or an old programmer who is still writing procedural code in C. Object-oriented programmers never debug their code—they write unit tests. My point here is that unit testing is a technique that completely replaces debugging. If debugging is required, the design is bad.

%ce%b1%cf%81%cf%87%ce%b5%ce%af%ce%bf-%ce%bb%ce%ae%cf%88%ce%b7%cf%82
The Revenant (2015) by Alejandro G. Iñárritu

Let’s say I’m a bad imperative procedural programmer, and this is my Java code:

class FileUtils {
  public static Iterable<String> readWords(File f) {
    String text = new String(
      Files.readAllBytes(Paths.get(f)),
      "UTF-8"
    );
    Set<String> words = new HashSet<>();
    for (String word : text.split(" ")) {
      words.add(word);
    }
    return words;
  }
}

This static utility method reads file content and then finds all the unique words in it. Pretty simple. However, if it doesn’t work, what do we do? Let’s say this is the file:

We know what we are,
but know not what we may be.

From it, we get this list of words:

"We"
"know"
"what"
"we"
"are,\n"
"but"
"not"
"may"
"be\n"

Now that doesn’t look right to me … so what is the next step? Either the file reading doesn’t work correctly or the split is broken. Let’s debug, right? Let’s give it a file through an input and go step by step, tracing and watching the variables. We’ll find the bug and fix it. But when a similar problem shows up, we’ll have to debug again! And that’s what unit testing is supposed to prevent.

We’re supposed to create a unit test once, in which the problem is reproduced. Then we fix the problem and make sure the test passes. That’s how we save our investments in problem solving. We won’t fix it again, because it won’t happen again. Our test will prevent it from happening.

If you perceive debugging to be faster and easier, think about the quality of your code

However, all this will work only if it’s easy to create a unit test. If it’s difficult, I’ll be too lazy to do it. I will just debug and fix the problem. In this particular example, creating a test is a rather expensive procedure. What I mean is the complexity of the unit test will be rather high. We have to create a temporary file, fill it with data, run the method, and check the results. To find out what’s going on and where the bug is, I’ll have to create a number of tests. To avoid code duplication, I’ll also have to create some supplementary utilities to help me create that temporary file and fill it with data. That’s a lot of work. Well, maybe not “a lot”, but way more than a few minutes of debugging.

Thus, if you perceive debugging to be faster and easier, think about the quality of your code. I bet it has a lot of opportunities for refactoring, just like the code from the example above. Here is how I would modify it. First of all, I would turn it into a class, because utility static methods are a bad practice:

class Words implements Iterable<String> {
  private final File file;
  Words(File src) {
    this.file = src;
  }
  @Override
  public Iterator<String> iterator() {
    String text = new String(
      Files.readAllBytes(Paths.get(this.file)),
      "UTF-8"
    );
    Set<String> words = new HashSet<>();
    for (String word : text.split(" ")) {
      words.add(word);
    }
    return words.iterator();
  }
}

It looks better already, but the complexity is still there. Next, I would break it down into smaller classes:

class Text {
  private final File file;
  Text(File src) {
    this.file = src;
  }
  @Override
  public String toString() {
    return new String(
      Files.readAllBytes(Paths.get(this.file)),
      "UTF-8"
    );
  }
}
class Words implements Iterable<String> {
  private final String text;
  Words(String txt) {
    this.text = txt;
  }
  @Override
  public Iterator<String> iterator() {
    Set<String> words = new HashSet<>();
    for (String word : this.text.split(" ")) {
      words.add(word);
    }
    return words.iterator();
  }
}

What do you think now? Writing a test for the Words class is a pretty trivial task:

import org.junit.Test;
import static org.hamcrest.MatcherAssert.*;
import static org.hamcrest.Matchers.*;
public class WordsTest {
  @Test
  public void parsesSimpleText() {
    assertThat(
      new Words("How are you?"),
      hasItems("How", "are", "you")
    );
  }
}

How much time did that take? Less than a minute. We don’t need to create a temporary file and load it with data, because class Words doesn’t do anything with files. It just parses the incoming string and finds the unique words in it. Now it’s easy to fix, since the test is small and we can easily create more tests; for example:

import org.junit.Test;
import static org.hamcrest.MatcherAssert.*;
import static org.hamcrest.Matchers.*;
public class WordsTest {
  @Test
  public void parsesSimpleText() {
    assertThat(
      new Words("How are you?"),
      hasItems("How", "are", "you")
    );
  }
  @Test
  public void parsesMultipleLines() {
    assertThat(
      new Words("first line\nsecond line\n"),
      hasItems("first", "second", "line")
    );
  }
}

My point is that debugging is necessary when the amount of time to write a unit test is significantly more than the time it takes to click those Trace-In/Trace-Out buttons. And it’s logical. We all are lazy and want fast and easy solutions. But debugging burns time and wastes energy. It helps us find problems but doesn’t help prevent them from reappearing.

Debugging is needed when our code is procedural and algorithmic—when the code is all about how the goal should be achieved instead of what the goal is. See the examples above again. The first static method is all about how we read the file, parse it, and find words. It’s even named readWords() (a verb). To the contrary, the second example is about what will be achieved. It’s either the Text of the file or Words of the text (both are nouns).

I believe there is no place for debugging in clean object-oriented programming. Only unit testing!

Reference: Are You Still Debugging? from our JCG partner Yegor Bugayenko at the About Programming blog.

Yegor Bugayenko

Yegor Bugayenko is an Oracle certified Java architect, CEO of Zerocracy, author of Elegant Objects book series about object-oriented programing, lead architect and founder of Cactoos, Takes, Rultor and Jcabi, and a big fan of test automation.
Subscribe
Notify of
guest

This site uses Akismet to reduce spam. Learn how your comment data is processed.

3 Comments
Oldest
Newest Most Voted
Inline Feedbacks
View all comments
Rod
Rod
7 years ago

“If debugging is required, the design is bad” … all blanket statements are incorrect.

lorenzo
lorenzo
7 years ago

the best thing of this article is the title.

Zsolt
Zsolt
7 years ago

The author is absolutely right, this is the key sentence:

“But debugging burns time and wastes energy. It helps us find problems but doesn’t help prevent them from reappearing.”

Back to top button