Java equal() and hashCode() based on different fields?

Question

Is there any situation where it makes sense for a class to implement its equals() and hashCode() methods using a different set of the class fields?

I'm asking because I am puzzled by the Netbeans equals() and hashCode() generator, where you are asked to choose the fields to include in each method separately. I always end up selecting the same fields for both methods, but is there a situation where this is not the correct choice?

Answer 1

Well, equals() must use all the fields used by hashCode(), as otherwise you could get different hash codes for equal objects. The reverse isn't true though - you could choose not to take account of one particular field when choosing the hash code. That way you could end up with the same hash code for two unequal objects which only differed by that "unused" field (as opposed to through natural collisions). You'd only want that in a situation where you knew collisions would be unlikely but where you were going to be hashing a lot. I imagine it's extremely rare :)

Another case would be where you had some sort of custom equality comparison - such as case insensitive string comparisons - where it's tricky or expensive to generate a hash code for the field. Again, this would lead to more likelihood of collision but would be valid.