When running FindBugs on my project, I got a few instances of the error described above.
Namely, my overriding versions of equals cast the RHS object into the same type as the object in which the overriding version is defined.
However, I'm not sure whether a better design is possible, since AFAIK Java does not allow variance in method parameters, so it is not possible to define any other type for the equals parameter.
Am I doing something very wrong, or is FindBugs too eager?
A different way to phrase this question is: what is the correct behavior if the object passed to equals is not the same type as an LHS: Is this a false, or should there be an exception?
For example:
public boolean equals(Object rhs)
{
MyType rhsMyType = (MyType)rhs; // Should throw exception
if(this.field1().equals(rhsMyType.field1())... // Or whatever
}
Typically, when implementing equals you can check to see whether the class of the argument is equal (or compatible) to the implementing class before casting it. Something like this:
if (getClass() != obj.getClass())
return false;
MyObj myObj = (MyObj) obj;
Doing it this way will prevent the FindBugs warning.
A side note to address a comment:
Some people argue to use instanceof
instead of getClass
to check type safety. There is a big debate on that, which I was trying not to get into when I noted that you can check for class equality or compatibility, but I guess I can't escape it. It boils down to this - if you use instanceof
you can support equality between instances of a class and instances of its subclass, but you risk breaking the symmetric contract of equals
. Generally I would recommend not to use instanceof
unless you know you need it and you know what you are doing. For more information see: