Skip to content
David Bürgin edited this page Nov 8, 2015 · 25 revisions

Checker Framework tutorial

Welcome!

This is a hands-on, play-along tutorial for null-safe programming with the Checker Framework.

Why?

Suppose you have created a neat little utility that turns the String A wonderful autumn afternoon into a-wonderful-autumn-afternoon:

public static String makeSlug(String s) {
    return s.replaceAll("\\s+", "-").toLowerCase();
}

Somewhere else in the app this utility is used to create a page title:

    String title = hasTitle()
            ? getTitle()
            : System.getProperty("defaults.title");

    String titleSlug = makeSlug(title);

What could go wrong?

If you have been programming Java for any amount of time, the call to makeSlug() probably has you shifting around uneasily: this code may blow up with a NullPointerException.

Though it doesn't make it explicit, makeSlug does not accept a null argument. Just from reading the code we can see that at the point where makeSlug(title) gets evaluated, title might very well be null. Therefore the code is not correct.

This is the kind of reasoning we do every day: read the code and track which references may be null at a given point to make sure the code is correct.

Wouldn't it be nice if this sort of reasoning about what may and may not be null at a given place in the code could somehow be expressed in code? And wouldn't it be nice to have a machine check the flow of ‘possibly null’ and ‘definitely not null’ references for us, and establish whether the code is correct with respect to null?

This is where the Checker Framework comes in.

The Checker Framework

The Checker Framework consists of a number of ‘checkers’ that hook into the compilation process. Each checker makes a promise to check something at compile time that would normally only be detected at run-time. There are checkers for nullness, for regexps, for format strings, and so on. For this to work, we need to make some of our assumptions explicit through annotations.

The Regex Checker, for example, checks the syntax and usage of Strings representing a regular expression.

@Regex String digits = "\\d+";

Provided that you have annotated your @Regexes properly, you get a compile time guarantee that no errors from mistaken use of regexps arise at runtime.

The checker we are going to use is the Nullness Checker, and its promise is to eliminate all errors like the NullPointerException we saw at the beginning.

This tutorial

This is an interactive tutorial for you to follow along, and I mean it.

  • We will check out the code, set up a project, and set up the Checker Framework.
  • The Checker will find errors and we will interpret the diagnostics, discuss, and fix some bugs.
  • The entire play-along part is encapsulated nicely in the Git log for reference.
  • I promise to give a summary of achievements (annotations added, bugs fixed) in the end. LINK

With that said, we can proceed to the mission statement on the next page.

Clone this wiki locally