With the advent of .NET 4.5.3, WPF developers now have three (or more) ways to notify the INotifyPropertyChanged
Interface of property changes. Basically, my question is Which of the two methods introduced from.NET 4.5 onwards is the more efficient way to notify property changes and whether either way has any benefit when being used in WPF?
Background
For those not so familiar with this subject, here are the main three methods. The first is the original, more error prone method of simply passing a string:
public string TestValue
{
get { return testValue; }
set { testValue = value; NotifyPropertyChanged("TestValue"); }
}
protected virtual void NotifyPropertyChanged(string propertyName)
{
if (PropertyChanged != null)
{
PropertyChanged(this, new PropertyChangedEventArgs(propertyName));
}
}
The second method was introduced in .NET 4.5; the CallerMemberNameAttribute
:
public string TestValue
{
get { return testValue; }
set { testValue = value; NotifyPropertyChanged(); }
}
protected virtual void NotifyPropertyChanged([CallerMemberName]string propertyName = "")
{
if (PropertyChanged != null)
{
PropertyChanged(this, new PropertyChangedEventArgs(propertyName));
}
}
The third and most recent method was (or will soon be) introduced in C#6.0 as part of .NET 4.5.3; the nameof
Operator:
public string TestValue
{
get { return testValue; }
set { testValue = value; NotifyPropertyChanged(nameof(TestValue)); }
}
protected virtual void NotifyPropertyChanged(string propertyName)
{
PropertyChanged?.Invoke(this, new PropertyChangedEventArgs(propertyName));
}
My own assumption would be that the original, more error prone method of simply passing a string would be the most efficient, as I can only imagine that the other two methods use some form of reflection. However, I'm really keen to find out which of the other two methods is more efficient and whether there would actually be any difference between using the CallerMemberNameAttribute
attribute and the nameof
operator in a WPF context.
About efficiency: using a string directly,
CallerMemberNameAttribute
,nameof
are all exactly the same since the string is injected by the compiler at compile-time. There's no reflection involved.We can see that using TryRoslyn that produces this for
CallerMemberNameAttribute
:And this for
nameof
:Since at runtime all options are simply a
string
there's no problem with the WPF context.About convenience:
CallerMemberNameAttribute
requires you to have an optional parameter whilenameof
doesn't butnameof
requires you to specify the property whileCallerMemberNameAttribute
doesn't.I predict that
nameof
would turn out to be so popular that it would be much simpler to use it instead.