AreNotSame uses reference equality (object.ReferenceEquals) - i.e. are they the same actual instance of an object; AreNotEqual uses conceptual equality (.Equals) - i.e. are they considered equal.
Almost all the answers given here are correct, but it's probably worth giving an example:
public static string GetSecondWord(string text)
{
// Yes, an appalling implementation...
return text.Split(' ')[1];
}
string expected = "world";
string actual = GetSecondWord("hello world");
// Good: the two strings should be *equal* as they have the same contents
Assert.AreEqual(expected, actual);
// Bad: the two string *references* won't be the same
Assert.AreSame(expected, actual);
AreNotEqual and AreNotSame are just inversions of AreEqual and AreSame of course.
If you use Assert.AreSame with value types, they are boxed. In other words, it's equivalent to doing:
int firstNumber = 1;
int secondNumber = 1;
object boxedFirstNumber = firstNumber;
object boxedSecondNumber = secondNumber;
// There are overloads for AreEqual for various value types
// (assuming NUnit here)
Assert.AreEqual(firstNumber, secondNumber);
// ... but not for AreSame, as it's not intended for use with value types
Assert.AreSame(boxedFirstNumber, boxedSecondNumber);
Neither firstNumber nor secondNumber has an object value, because int is a value type. The reason the AreSame call will fail is because in .NET, boxing a value creates a new box each time. (In Java it sometimes doesn't - this has caught me out before.)
Basically you should never use AreSame when comparing value types. When you're comparing reference types, use AreSame if you want to check for identical references; use AreEqual to check for equivalence under Equals. EDIT: Note that there are situations where NUnit doesn't just use Equals directly; it has built-in support for collections, where the elements in the collections are tested for equality.
The claim in the answer that:
Using the example above changing the
int to string, AreSame and AreEqual
will return the same value.
entirely depends on how the variables are initialized. If they use string literals, then yet, interning will take care of that. If, however, you use:
then AreSame and AreEqual will almost certainly not return the same value.
As for:
The general rule of thumb is to use
AreEqual on value types and AreSame on
reference types.
I almost never want to check for reference identity. It's rarely useful to me. I want to check for equivalence which is what AreEqual checks for. (I'm not saying that AreSame shouldn't be there - it's a useful method, just far more rarely than AreEqual.)
Two things can be equal, but different objects. AreNotEqual checks the objects values via the equality test, while AreNotSame checks that they are not the same exact object.
It is obvious why we would want to test that things AreNotEqual (we care about the values being tested); what about AreNotSame? The usefulness of this in testing is found when you have passed references around and want to make sure that after your shuffling is done that two references are still the same object.
In a real world case, we use a lot of caching objects to mitigate round trips to the database. After an object has been handed off to the cache system, our unit tests ensure that in some cases we get back the same object (cache was valid) and in other cases we get back a fresh object (cache was invalidated). Note that AreNotEqual would not necessary suffice in this case. If the object had a new timestamp in the database, yet the data was not "different enough" to fail an equality test, AreNotEqual wouldn't notice that we refreshed the object.
Isn't it so that AreNotEqual checks for the case where two objects are not equal in terms of Equals() method, whereas AreNotSame checks for the case where the two object references are not the same. So if x and y are two objects which are equal in terms of Equals() but have been separately allocated, AreNotEqual() would trigger failing assertion but the other not.
Assert.AreNotEqual asserts that two values are not equal to each other.
Assert.AreNotSame asserts that two variables do not point to the same object.
Example 1:
Example 2:
AreNotSame uses reference equality (
object.ReferenceEquals
) - i.e. are they the same actual instance of an object; AreNotEqual uses conceptual equality (.Equals
) - i.e. are they considered equal.Almost all the answers given here are correct, but it's probably worth giving an example:
AreNotEqual
andAreNotSame
are just inversions ofAreEqual
andAreSame
of course.EDIT: A rebuttal to the currently accepted answer...
If you use
Assert.AreSame
with value types, they are boxed. In other words, it's equivalent to doing:Neither
firstNumber
norsecondNumber
has an object value, becauseint
is a value type. The reason theAreSame
call will fail is because in .NET, boxing a value creates a new box each time. (In Java it sometimes doesn't - this has caught me out before.)Basically you should never use
AreSame
when comparing value types. When you're comparing reference types, useAreSame
if you want to check for identical references; useAreEqual
to check for equivalence underEquals
. EDIT: Note that there are situations where NUnit doesn't just useEquals
directly; it has built-in support for collections, where the elements in the collections are tested for equality.The claim in the answer that:
entirely depends on how the variables are initialized. If they use string literals, then yet, interning will take care of that. If, however, you use:
then AreSame and AreEqual will almost certainly not return the same value.
As for:
I almost never want to check for reference identity. It's rarely useful to me. I want to check for equivalence which is what
AreEqual
checks for. (I'm not saying thatAreSame
shouldn't be there - it's a useful method, just far more rarely thanAreEqual
.)Two things can be equal, but different objects. AreNotEqual checks the objects values via the equality test, while AreNotSame checks that they are not the same exact object.
It is obvious why we would want to test that things AreNotEqual (we care about the values being tested); what about AreNotSame? The usefulness of this in testing is found when you have passed references around and want to make sure that after your shuffling is done that two references are still the same object.
In a real world case, we use a lot of caching objects to mitigate round trips to the database. After an object has been handed off to the cache system, our unit tests ensure that in some cases we get back the same object (cache was valid) and in other cases we get back a fresh object (cache was invalidated). Note that AreNotEqual would not necessary suffice in this case. If the object had a new timestamp in the database, yet the data was not "different enough" to fail an equality test, AreNotEqual wouldn't notice that we refreshed the object.
Isn't it so that AreNotEqual checks for the case where two objects are not equal in terms of Equals() method, whereas AreNotSame checks for the case where the two object references are not the same. So if x and y are two objects which are equal in terms of Equals() but have been separately allocated, AreNotEqual() would trigger failing assertion but the other not.
AreNotSame does reference comparison, whereas AreNotEqual does an equality comparison.