OnlyOneReturn Considered Harmful

WARNING: This blog entry was imported from my old blog on blogs.sun.com (which used different blogging software), so formatting and links may not be correct.

The other day I

mentioned PMD
and how it's vital to learn how to customize the ruleset, since by default
it includes some rules that are evil. Evil I tell you!

My favorite rule to hate is the OnlyOneReturn rule. This rule basically
says that

A method should have only one exit point, and that should be the last statement in the method.

Unfortunately, the rule author does not provide supporting documentation
for this point of view, and I personally cannot think of any.
In fact, restricting yourself to a single return statement makes the code
less readable. Take the following method from the JDK's Integer
class for example.

public static String toString(int i) {
if (i == Integer.MIN_VALUE) {
return "-2147483648";
}

int size = (i < 0) ? (stringSize(-i) + 1) : stringSize(i);
char[] buf = new char[size];
getChars(i, size, buf);

return new String(0, size, buf);
}

We have a special case, and it's dealt with right at the beginning of
the method. If we hadn't done that, we would need to slap a big else
block around the remainder of the code, as well as introduce a temporary
variable, e.g. result, to hold the return value that we then
return as the last statement of the method.

The above is a trivial example, but I frequently write code where I have
multiple exit points early in the method that deal with special cases, and
this avoids the need for deeply nested code.

Creator's Visual Page Designer Source Code: Proud OnlyOneReturn Rule Violator
Since 2003.