Question

I have a question with whether or not this is a standard for using IComparer in C#. Say I have a situation in which there are three Person objects: P1, P2, and P3. Say I call the Compare method passing in P1 and P2 and the result is 0. This essentially means the two people should be categorized as equal. Now say I call the Compare method passing in P2 and P3 and the result for that is 0 as well. Again, this means the two people are equal. Logically speaking, one can assume P1 and P3 are equal as well; however, the Compare method could be implemented however someone decides to implement it. So is it a standard to implement it in such a way that P1 and P3 would also return 0 in this case?

Here’s code of what I’m asking:

// Assume these are initialized properly
Person p1 = null, p2 = null, p3 = null;
IComparer<Person> comparer = null;

// Compare person 1 to person 2 and result is 0
Debug.Assert(comparer.Compare(p1, p2) == 0);

// Compare person 2 to person 3 and result is 0
Debug.Assert(comparer.Compare(p2, p3) == 0);

// Would this be a fair assumption that person 1 and person 3 would also be 0?
Debug.Assert(comparer.Compare(p1, p3) == 0);

Answer 1

Yes, that would be the standard. Its explicitly stated for IComparable:

If A.CompareTo(B) returns zero and
B.CompareTo(C) returns zero, then
A.CompareTo(C) is required to return
zero.

I can’t find anything in the official documentation that comes right out and states the same thing for ICompare but I think its safe to assume the same holds true.