For example, suppose I want an ICar
interface and that all implementations will contain the field Year
. Does this mean that every implementation has to separately declare Year
? Wouldn't it be nicer to simply define this in the interface?
C# – Why can’t C# interfaces contain fields
cinterface
Related Solutions
Java 8 permits static interface methods
With Java 8, interfaces can have static methods. They can also have concrete instance methods, but not instance fields.
There are really two questions here:
- Why, in the bad old days, couldn't interfaces contain static methods?
- Why can't static methods be overridden?
Static methods in interfaces
There was no strong technical reason why interfaces couldn't have had static methods in previous versions. This is summed up nicely by the poster of a duplicate question. Static interface methods were initially considered as a small language change, and then there was an official proposal to add them in Java 7, but it was later dropped due to unforeseen complications.
Finally, Java 8 introduced static interface methods, as well as override-able instance methods with a default implementation. They still can't have instance fields though. These features are part of the lambda expression support, and you can read more about them in Part H of JSR 335.
Overriding static methods
The answer to the second question is a little more complicated.
Static methods are resolvable at compile time. Dynamic dispatch makes sense for instance methods, where the compiler can't determine the concrete type of the object, and, thus, can't resolve the method to invoke. But invoking a static method requires a class, and since that class is known statically—at compile time—dynamic dispatch is unnecessary.
A little background on how instance methods work is necessary to understand what's going on here. I'm sure the actual implementation is quite different, but let me explain my notion of method dispatch, which models observed behavior accurately.
Pretend that each class has a hash table that maps method signatures (name and parameter types) to an actual chunk of code to implement the method. When the virtual machine attempts to invoke a method on an instance, it queries the object for its class and looks up the requested signature in the class's table. If a method body is found, it is invoked. Otherwise, the parent class of the class is obtained, and the lookup is repeated there. This proceeds until the method is found, or there are no more parent classes—which results in a NoSuchMethodError
.
If a superclass and a subclass both have an entry in their tables for the same method signature, the sub class's version is encountered first, and the superclass's version is never used—this is an "override".
Now, suppose we skip the object instance and just start with a subclass. The resolution could proceed as above, giving you a sort of "overridable" static method. The resolution can all happen at compile-time, however, since the compiler is starting from a known class, rather than waiting until runtime to query an object of an unspecified type for its class. There is no point in "overriding" a static method since one can always specify the class that contains the desired version.
Constructor "interfaces"
Here's a little more material to address the recent edit to the question.
It sounds like you want to effectively mandate a constructor-like method for each implementation of IXMLizable
. Forget about trying to enforce this with an interface for a minute, and pretend that you have some classes that meet this requirement. How would you use it?
class Foo implements IXMLizable<Foo> {
public static Foo newInstanceFromXML(Element e) { ... }
}
Foo obj = Foo.newInstanceFromXML(e);
Since you have to explicitly name the concrete type Foo
when "constructing" the new object, the compiler can verify that it does indeed have the necessary factory method. And if it doesn't, so what? If I can implement an IXMLizable
that lacks the "constructor", and I create an instance and pass it to your code, it is an IXMLizable
with all the necessary interface.
Construction is part of the implementation, not the interface. Any code that works successfully with the interface doesn't care about the constructor. Any code that cares about the constructor needs to know the concrete type anyway, and the interface can be ignored.
The point of Dispose is to free unmanaged resources. It needs to be done at some point, otherwise they will never be cleaned up. The garbage collector doesn't know how to call DeleteHandle()
on a variable of type IntPtr
, it doesn't know whether or not it needs to call DeleteHandle()
.
Note: What is an unmanaged resource? If you found it in the Microsoft .NET Framework: it's managed. If you went poking around MSDN yourself, it's unmanaged. Anything you've used P/Invoke calls to get outside of the nice comfy world of everything available to you in the .NET Framework is unmanaged – and you're now responsible for cleaning it up.
The object that you've created needs to expose some method, that the outside world can call, in order to clean up unmanaged resources. The method can be named whatever you like:
public void Cleanup()
or
public void Shutdown()
But instead there is a standardized name for this method:
public void Dispose()
There was even an interface created, IDisposable
, that has just that one method:
public interface IDisposable
{
void Dispose()
}
So you make your object expose the IDisposable
interface, and that way you promise that you've written that single method to clean up your unmanaged resources:
public void Dispose()
{
Win32.DestroyHandle(this.CursorFileBitmapIconServiceHandle);
}
And you're done. Except you can do better.
What if your object has allocated a 250MB System.Drawing.Bitmap (i.e. the .NET managed Bitmap class) as some sort of frame buffer? Sure, this is a managed .NET object, and the garbage collector will free it. But do you really want to leave 250MB of memory just sitting there – waiting for the garbage collector to eventually come along and free it? What if there's an open database connection? Surely we don't want that connection sitting open, waiting for the GC to finalize the object.
If the user has called Dispose()
(meaning they no longer plan to use the object) why not get rid of those wasteful bitmaps and database connections?
So now we will:
- get rid of unmanaged resources (because we have to), and
- get rid of managed resources (because we want to be helpful)
So let's update our Dispose()
method to get rid of those managed objects:
public void Dispose()
{
//Free unmanaged resources
Win32.DestroyHandle(this.CursorFileBitmapIconServiceHandle);
//Free managed resources too
if (this.databaseConnection != null)
{
this.databaseConnection.Dispose();
this.databaseConnection = null;
}
if (this.frameBufferImage != null)
{
this.frameBufferImage.Dispose();
this.frameBufferImage = null;
}
}
And all is good, except you can do better!
What if the person forgot to call Dispose()
on your object? Then they would leak some unmanaged resources!
Note: They won't leak managed resources, because eventually the garbage collector is going to run, on a background thread, and free the memory associated with any unused objects. This will include your object, and any managed objects you use (e.g. the
Bitmap
and theDbConnection
).
If the person forgot to call Dispose()
, we can still save their bacon! We still have a way to call it for them: when the garbage collector finally gets around to freeing (i.e. finalizing) our object.
Note: The garbage collector will eventually free all managed objects. When it does, it calls the
Finalize
method on the object. The GC doesn't know, or care, about your Dispose method. That was just a name we chose for a method we call when we want to get rid of unmanaged stuff.
The destruction of our object by the Garbage collector is the perfect time to free those pesky unmanaged resources. We do this by overriding the Finalize()
method.
Note: In C#, you don't explicitly override the
Finalize()
method. You write a method that looks like a C++ destructor, and the compiler takes that to be your implementation of theFinalize()
method:
~MyObject()
{
//we're being finalized (i.e. destroyed), call Dispose in case the user forgot to
Dispose(); //<--Warning: subtle bug! Keep reading!
}
But there's a bug in that code. You see, the garbage collector runs on a background thread; you don't know the order in which two objects are destroyed. It is entirely possible that in your Dispose()
code, the managed object you're trying to get rid of (because you wanted to be helpful) is no longer there:
public void Dispose()
{
//Free unmanaged resources
Win32.DestroyHandle(this.gdiCursorBitmapStreamFileHandle);
//Free managed resources too
if (this.databaseConnection != null)
{
this.databaseConnection.Dispose(); //<-- crash, GC already destroyed it
this.databaseConnection = null;
}
if (this.frameBufferImage != null)
{
this.frameBufferImage.Dispose(); //<-- crash, GC already destroyed it
this.frameBufferImage = null;
}
}
So what you need is a way for Finalize()
to tell Dispose()
that it should not touch any managed resources (because they might not be there anymore), while still freeing unmanaged resources.
The standard pattern to do this is to have Finalize()
and Dispose()
both call a third(!) method; where you pass a Boolean saying if you're calling it from Dispose()
(as opposed to Finalize()
), meaning it's safe to free managed resources.
This internal method could be given some arbitrary name like "CoreDispose", or "MyInternalDispose", but is tradition to call it Dispose(Boolean)
:
protected void Dispose(Boolean disposing)
But a more helpful parameter name might be:
protected void Dispose(Boolean itIsSafeToAlsoFreeManagedObjects)
{
//Free unmanaged resources
Win32.DestroyHandle(this.CursorFileBitmapIconServiceHandle);
//Free managed resources too, but only if I'm being called from Dispose
//(If I'm being called from Finalize then the objects might not exist
//anymore
if (itIsSafeToAlsoFreeManagedObjects)
{
if (this.databaseConnection != null)
{
this.databaseConnection.Dispose();
this.databaseConnection = null;
}
if (this.frameBufferImage != null)
{
this.frameBufferImage.Dispose();
this.frameBufferImage = null;
}
}
}
And you change your implementation of the IDisposable.Dispose()
method to:
public void Dispose()
{
Dispose(true); //I am calling you from Dispose, it's safe
}
and your finalizer to:
~MyObject()
{
Dispose(false); //I am *not* calling you from Dispose, it's *not* safe
}
Note: If your object descends from an object that implements
Dispose
, then don't forget to call their base Dispose method when you override Dispose:
public override void Dispose()
{
try
{
Dispose(true); //true: safe to free managed resources
}
finally
{
base.Dispose();
}
}
And all is good, except you can do better!
If the user calls Dispose()
on your object, then everything has been cleaned up. Later on, when the garbage collector comes along and calls Finalize, it will then call Dispose
again.
Not only is this wasteful, but if your object has junk references to objects you already disposed of from the last call to Dispose()
, you'll try to dispose them again!
You'll notice in my code I was careful to remove references to objects that I've disposed, so I don't try to call Dispose
on a junk object reference. But that didn't stop a subtle bug from creeping in.
When the user calls Dispose()
: the handle CursorFileBitmapIconServiceHandle is destroyed. Later when the garbage collector runs, it will try to destroy the same handle again.
protected void Dispose(Boolean iAmBeingCalledFromDisposeAndNotFinalize)
{
//Free unmanaged resources
Win32.DestroyHandle(this.CursorFileBitmapIconServiceHandle); //<--double destroy
...
}
The way you fix this is tell the garbage collector that it doesn't need to bother finalizing the object – its resources have already been cleaned up, and no more work is needed. You do this by calling GC.SuppressFinalize()
in the Dispose()
method:
public void Dispose()
{
Dispose(true); //I am calling you from Dispose, it's safe
GC.SuppressFinalize(this); //Hey, GC: don't bother calling finalize later
}
Now that the user has called Dispose()
, we have:
- freed unmanaged resources
- freed managed resources
There's no point in the GC running the finalizer – everything's taken care of.
Couldn't I use Finalize to cleanup unmanaged resources?
The documentation for Object.Finalize
says:
The Finalize method is used to perform cleanup operations on unmanaged resources held by the current object before the object is destroyed.
But the MSDN documentation also says, for IDisposable.Dispose
:
Performs application-defined tasks associated with freeing, releasing, or resetting unmanaged resources.
So which is it? Which one is the place for me to cleanup unmanaged resources? The answer is:
It's your choice! But choose
Dispose
.
You certainly could place your unmanaged cleanup in the finalizer:
~MyObject()
{
//Free unmanaged resources
Win32.DestroyHandle(this.CursorFileBitmapIconServiceHandle);
//A C# destructor automatically calls the destructor of its base class.
}
The problem with that is you have no idea when the garbage collector will get around to finalizing your object. Your un-managed, un-needed, un-used native resources will stick around until the garbage collector eventually runs. Then it will call your finalizer method; cleaning up unmanaged resources. The documentation of Object.Finalize points this out:
The exact time when the finalizer executes is undefined. To ensure deterministic release of resources for instances of your class, implement a Close method or provide a
IDisposable.Dispose
implementation.
This is the virtue of using Dispose
to cleanup unmanaged resources; you get to know, and control, when unmanaged resource are cleaned up. Their destruction is "deterministic".
To answer your original question: Why not release memory now, rather than for when the GC decides to do it? I have a facial recognition software that needs to get rid of 530 MB of internal images now, since they're no longer needed. When we don't: the machine grinds to a swapping halt.
Bonus Reading
For anyone who likes the style of this answer (explaining the why, so the how becomes obvious), I suggest you read Chapter One of Don Box's Essential COM:
- Direct link: Chapter 1 sample by Pearson Publishing
- magnet: 84bf0b960936d677190a2be355858e80ef7542c0
In 35 pages he explains the problems of using binary objects, and invents COM before your eyes. Once you realize the why of COM, the remaining 300 pages are obvious, and just detail Microsoft's implementation.
I think every programmer who has ever dealt with objects or COM should, at the very least, read the first chapter. It is the best explanation of anything ever.
Extra Bonus Reading
When everything you know is wrong archiveby Eric Lippert
It is therefore very difficult indeed to write a correct finalizer, and the best advice I can give you is to not try.
Related Topic
- C# – Interface defining a constructor signature
- Interface vs Abstract Class (general OO)
- C# – Pass Method as Parameter using C#
- C# – Why are C# 4 optional parameters defined on interface not enforced on implementing class
- C# – Entity Framework 5 Updating a Record
- Scala: list.flatten: no implicit argument matching parameter type (Any) = > Iterable[Any] was found
- Sql – Prepared statements and the IN expression
Best Answer
Though many of the other answers are correct at the semantic level, I find it interesting to also approach these sorts of questions from the implementation details level.
An interface can be thought of as a collection of slots, which contain methods. When a class implements an interface, the class is required to tell the runtime how to fill in all the required slots. When you say
the class says "when you create an instance of me, stuff a reference to Foo.M in the slot for IFoo.M.
Then when you do a call:
the compiler generates code that says "ask the object what method is in the slot for IFoo.M, and call that method.
If an interface is a collection of slots that contain methods, then some of those slots can also contain the get and set methods of a property, the get and set methods of an indexer, and the add and remove methods of an event. But a field is not a method. There's no "slot" associated with a field that you can then "fill in" with a reference to the field location. And therefore, interfaces can define methods, properties, indexers and events, but not fields.