Clojure's "zero?" has some quirks

In short, Clojure’s zero? isn’t exactly like nil? and true? and others. It checks different numeric types and throws if given a non-numeric type.

This post was written about Clojure 1.8.0.

Clojure has a few functions that do equality checks, like nil?, true?, false?, and zero?. The docs all look about the same for these functions. nil? “returns true if x is nil, false otherwise” and the rest look similar.

But despite similar documentation, zero? is a little bit different from the rest.

Let’s start by looking at true?. It accepts anything as an argument; a boolean, a string, whatever.

(true? true)   ;; true
(true? false)  ;; false
(true? "str")  ;; false

In contrast, zero? throws an error if the argument isn’t a number.

(zero? 0)      ;; true
(zero? 42.0)   ;; false
(zero? "str")  ;; throws java.lang.ClassCastException

This was a little surprising to me! It’s not documented and it also doesn’t match the other equality checkers.

zero? is also a little smarter about different numeric types. zero? handles cases that a simple equality check doesn’t.

(= 0 0)      ;; true
(= 0 0.0)    ;; false
(= 0 "str")  ;; false

(zero? 0)    ;; true
(zero? 0.0)  ;; true
(zero? "0")  ;; throws java.lang.ClassCastException

You could write a version of zero? that works like nil?, which won’t throw on non-numeric input but will still work for various number types.

(defn safe-zero? [x]
  (and (number? x) (zero? x)))

(safe-zero? 0)      ;; true
(safe-zero? 0.0)    ;; true
(safe-zero? "str")  ;; false

zero? isn’t vastly more complicated than I expected, but it’s a little more nuanced than I thought at first glance.