HashCode on implementation of Set

Question

javadoc on Set<E> hashcode:
"int hashCode()
"Returns the hash code value for this set. The hash code of a set is defined to be the sum of the hash codes of the elements in the set, where the hashcode of a null element is defined to be zero. This ensures that s1.equals(s2) implies that s1.hashCode()==s2.hashCode() for any two sets s1 and s2, as required by the general contract of the Object.hashCode method."
Let's say I have a class Foo which is an implementation of Set<Bar>. Bar's hashCode is a perfect hash. But the sum of such hash codes won't be a perfect hash on Foo instances, and I'd like to have Foo's hashCode return a perfect hash. (A perfect hash is possible given the domain of Bar instances). If Foo's hashCode returns a value which meets the assurance stated in the doc, but is not the sum of the element hash codes, do I cause a potential client problem by violating what the hash code is "defined to be"?

Answer 1

In addition to the excelent advice jverd gave I'd
like to ask if beeing a Set<Bar> is Foo's primary
job. Do clients view it as anything else, than a
Set<Bar>? (If it's usually referenced from a
variabled delcard "Foo foo;" and not "Set<Bar> foo",
then chances are, that beeing a Set is not it's
primary job).Yes, being a Set<Bar> is its primary, and probably only, job. The name of the class is (the equivalent of) BarSet.
>
If indeed Foo "is" "not just" a Set<Bar>, then you
might want to remove that interface from it and
produce a delegate, that acts as the Set.
(foo.getSet())
This way you could implement hashCode() of Foo the
perfect way you want it to and have
foo.getSet().hashCode() follow the requirements laid
out by Set.hashCode()).
This way you could fullfill both requirements.Or, as kajbj implies, I could fulfill both by just supplying a perfectHashCode(). I like that idea.
Message was edited by:
com.stevebrecher to correct misspelling of a name