Are these two lines the same, '? ... :' vs '??'?
Is there a difference between these two lines?
MyName = (s.MyName == null) ? string.Empty : s.MyName
or
MyName = s.MyName ?? string.Empty
UPDATE: I wrote a blog post that discusses this topic in more depth. http://www.codeducky.org/properties-fields-and-methods-oh-my/
Generally they will return the same result. However, there are a few cases where you will experience noticeable differences when MyName
is a property because the MyName
getter will be executed twice in the first example and only once in the second example.
For example, you may experience performance differences from executing MyName
twice:
string MyName
{
get
{
Thread.Sleep(10000);
return "HELLO";
}
}
Or you may get different results from executing MyName
twice if MyName
is stateful:
private bool _MyNameHasBeenRead = false;
string MyName
{
get
{
if(_MyNameHasBeenRead)
throw new Exception("Can't read MyName twice");
_MyNameHasBeenRead = true;
Thread.Sleep(10000);
return "HELLO";
}
}
Or you may get different results from executing MyName
twice if MyName
can be changed on a different thread:
void ChangeMyNameAsync()
{
//MyName set to null in another thread which makes it
//possible for the first example to return null
Task.Run(() => this.MyName = null);
}
string MyName { get; set; }
Here's how the actual code is compiled. First the piece with the ternary expression:
IL_0007: ldloc.0 // s
IL_0008: callvirt s.get_MyName <-- first call
IL_000D: brfalse.s IL_0017
IL_000F: ldloc.0 // s
IL_0010: callvirt s.get_MyName <-- second call
IL_0015: br.s IL_001C
IL_0017: ldsfld System.String.Empty
IL_001C: call set_MyName
and here is the piece with the null-coalescing operator:
IL_0007: ldloc.0 // s
IL_0008: callvirt s.get_MyName <-- only call
IL_000D: dup
IL_000E: brtrue.s IL_0016
IL_0010: pop
IL_0011: ldsfld System.String.Empty
IL_0016: call s.set_MyName
As you can see the compiled code for the ternary operator will make two calls to get the property value, whereas the null-coalescing operator will only do 1.
If the property is more than a simple getter, you might be executing a function twice in the non-null case for the first one.
If the property is in a stateful object, then the second call to the property might return a different result:
class MyClass
{
private IEnumerator<string> _next = Next();
public MyClass()
{
this._next.MoveNext();
}
public string MyName
{
get
{
var n = this._next.Current;
this._next.MoveNext();
return n;
}
}
public static IEnumerator<string> Next()
{
yield return "foo";
yield return "bar";
}
}
Also, in the non-string case, the class might overload == to do something different than the ternary operator. I don't believe that the ternary operator can be overloaded.
The only difference is whether you evaluate s.MyName
twice or once. The first will do it twice in the case that s.MyName
is not null, the second will only ever evaluate it once.
In most cases, this difference doesn't matter, and I'd go with the second because it's more clear and concise.
Yes, both are the same, and it is the null-coalescing operator.
It returns the left-hand operand if the operand is not null; otherwise it returns the right hand operand.
If we talk about efficiency then
string MyName = (s.MyName == null) ? string.Empty : s.MyName;
string MyName2 = s.MyName ?? string.Empty;
If I use a dissembler then I can see that the first statement needs 19 statements to be executed by the compiler whereas the second statement required only 12 statements to be executed.