Changing values in HashSet

Question

I've read this question: Changing the elements in a set changes the 'equals' semantics

However, I don't know how to solve the problem that I can't change an item in the HashSet and remove it later.

I have some example sourcecode:

public static void main(String[] args) {
    TestClass testElement = new TestClass("1");
    Set<TestClass> set = new HashSet<>();
    set.add(testElement);
    printIt(testElement, set, "First Set");
    testElement.setS1("asdf");
    printIt(testElement, set, "Set after changing value");
    set.remove(testElement);
    printIt(testElement, set, "Set after trying to remove value");
    testElement.setS1("1");
    printIt(testElement, set, "Set after changing value back");
    set.remove(testElement);
    printIt(testElement, set, "Set removing value");
}

private static void printIt(TestClass hullo, Set<TestClass> set, String message) {
    System.out.println(message + " (hashCode is " + hullo.hashCode() + "):");
    for (TestClass testClass : set) {
        System.out.println("    " + testClass.toString());
        System.out.println("        HashCode: " + testClass.hashCode());
        System.out.println("        Element is equal: " + hullo.equals(testClass));
    }
}

Where TestClass is just a POJO that holds a variable (plus getter & setter) and has hashcode() and equals() implemented.

There was a request to show the equals() and hashcode()-methods. These are autogenerated by eclipse:

@Override
public int hashCode() {
    final int prime = 31;
    int result = 1;
    result = prime * result + ((s1 == null) ? 0 : s1.hashCode());
    return result;
}

@Override
public boolean equals(Object obj) {
    if (this == obj)
        return true;
    if (obj == null)
        return false;
    if (getClass() != obj.getClass())
        return false;
    TestClass other = (TestClass) obj;
    if (s1 == null) {
        if (other.s1 != null)
            return false;
    } else if (!s1.equals(other.s1))
        return false;
    return true;
}

The result is the following:

First Set (hashCode is 80):
    TestClass [s1=1]
        HashCode: 80
        Element is equal: true
Set after changing value (hashCode is 3003475):
    TestClass [s1=asdf]
        HashCode: 3003475
        Element is equal: true
Set after trying to remove value (hashCode is 3003475):
    TestClass [s1=asdf]
        HashCode: 3003475
        Element is equal: true
Set after changing value back (hashCode is 80):
    TestClass [s1=1]
        HashCode: 80
        Element is equal: true
Set removing value (hashCode is 80):

When the hashcode has changed, I can't remove the value from the HashSet. As in the linked question, I understand why it is like that, but I don't know how to delete a changed value. Is there any possibility to do so?

Could you post the hashcode and equals methods please? – M. Abbas Jun 20 '13 at 09:43 — M. Abbas, Jun 20 '13 at 09:43

Juned Ahsan · Accepted Answer · 2013-06-20T11:01:16.857

8

You are facing the problem because the keys in your hashset are not immutable. If you don't have immutable keys, you will lose the reference of the original key object once modified. And will never be able to get handle of that, which is sometimes referred as memory leaks in collection. So if you use immutable keys, you wouldn't run into this situation.

edited Jun 20 '13 at 11:01

answered Jun 20 '13 at 09:23

Juned Ahsan

67,789
12
98
136

Interesting - is that the result of the hash set actually being a wrapper around a hash map? – robjohncox Jun 20 '13 at 13:53
@robjohncox It is because the way hashing works. Hashcodes are used to store and retrieve objects. Suppose you create a key object and when you put it in a hashmap, its hashcode method will be called to caluculate the hash and find the bucket to store in. When you try to retrieve it, again hashcode is called to get the hash/bucket where key is stored. If you change the key object after storing in the set/map then the hashcode method will return a different hash for that key object, which is not the same as was used to store the key. – Juned Ahsan Jun 20 '13 at 13:59
1

Very good but to search reference in HashSet(hashmap) not only hashcode is used but also equals is used – Volodymyr Levytskyi Jul 27 '13 at 08:53
@VolodymyrLevytskyi Undoubtedly! – Juned Ahsan Jul 27 '13 at 09:51
@JunedAhsan I agree you won't have the Object itself to index into the Set, like with the Set methods `contains` and `remove`. You say "will never be able to get handle of that" - but can't we just iterate through the Set? We can get all the members in the Set this way, no? – flow2k Sep 05 '17 at 08:07

score 2 · Answer 2 · answered Jun 20 '13 at 14:04

As the question you linked to details, and as others have pointed out, you're encountering the mutable key issue. I'll requote from the Javadoc:

Note: Great care must be exercised if mutable objects are used as set elements. The behavior of a set is not specified if the value of an object is changed in a manner that affects equals comparisons while the object is an element in the set.

As you pointed out, you get that. The question is, how do you actually remove the object given that's the case? You can't use Set.remove(), because your object is lost in the hash table. You can, however, use an Iterator to do it. Something like the following:

TestClass toRemove = <the same instance, but mutated>;
for (Iterator<TestClass> iter = set.iterator(); iter.hasNext(); ) {
  TestClass item = iter.next();
  if (toRemove.equals(item)) {
    iter.remove();
  }
}

This approach relies on the fact that standard equals() methods, like you're using, have an instance check, and that check will return true.

Please keep in mind this is not the right way to solve this problem. The right way is to either use an immutable key or "exercise great care", but it is a way to remove a mutated object from a HashSet.

M. Abbas · Answer 3 · 2013-06-20T13:51:23.733

When you add testElement to the HashSet, it selects a bucket based on the hash code for testElement. When you ask the HashSet if it contains a TestElement, it computes the hash code of the object that it's looking for and searches only in that bucket.

As your hashCode() is based on non-final field, the hash code can change behind the scenes of the HashSet. Thus completely invalidating the basic assumption of the HashSet.

A correct implementation for the Testclass would have the s1 field as final.

Changing values in HashSet

3 Answers3

Linked