Question

I am porting Java code in .NET and I am stuck in the following lines that (behave unexpectedly in .NET).

Java:

Map<Set<State>, Set<State>> sets = new HashMap<Set<State>, Set<State>>();

Set<State> p = new HashSet<State>();

if (!sets.containsKey(p)) { ... }

The equivalent .NET code could possibly be:

IDictionary<HashSet<State>, HashSet<State>> sets = new Dictionary<HashSet<State>, HashSet<State>>();

HashSet<State> p = new HashSet<State>();

 if (!sets.containsKey(p)) { /* (Add to a list). Always get here in .NET (??) */ }

However the code comparison fails, the program think that “sets” never contain Key “p” and eventually results in OutOfMemoryException.

Perhaps I am missing something, object equality and identity might be different between Java and .NET.

I tried implementing IComparable and IEquatable in class State but the results were the same.

Edit:

What the code does is: If the sets does not contain key “p” (which is a HashSet) it is going to add “p” at the end of a LinkedList>.

The State class (Java) is a simple class defined as:

public class State implements Comparable<State> {

boolean accept;
Set<Transition> transitions;

int number;
int id;

// ...

public int compareTo(State s) {
    return s.id - id;
}

public boolean equals(Object obj) {
    return super.equals(obj);
}

public int hashCode() {
    return super.hashCode();
}

Answer 1

In Java, the list and set implementations override equals etc. The equivalent doesn’t happen in .NET. However, you can create a dictionary using an appropriate equality comparer using HashSet<T>.CreateSetComparer and pass it to the Dictionary constructor like this:

IDictionary<HashSet<State>, HashSet<State>> sets =
    new Dictionary<HashSet<State>, HashSet<State>>
       (HashSet<State>.CreateSetComparer());

This is assuming that State overrides Equals/GetHashCode and ideally implements IEquatable<State> too.