When is Dispose necessary?

Joan Venge picture Joan Venge · Jul 30, 2009 · Viewed 9k times · Source

When you have code like:

Bitmap bmp = new Bitmap ( 100, 100 );
Graphics g = Graphics.FromImage ( bmp );

Pen p = new Pen ( Color.FromArgb ( 128, Color.Blue ), 1 );
Brush b = new SolidBrush ( Color.FromArgb ( 128, Color.Blue ) );

g.FillEllipse ( b, 0, 0, 99, 99 );    
g.FillRegion ( b, pictureBox1.Region );

pictureBox1.BackColor = Color.Transparent;
pictureBox1.Image = bmp;

Do you have to dispose the pen and brush? What about bmp and the g?

My main question is, if these were to be disposed manually, why don't they get disposed as soon as they get out of the scope? Is that what would happen, if you didn't dispose them manually? Is it the delay that makes people do this manually?

Answer

Pavel Minaev picture Pavel Minaev · Jul 30, 2009

Yes, you have to dispose them - not just pen and brush, but also Bitmap and Graphics.

They don't get disposed when they're out of scope because the variables themselves are references, not objects, and C# compiler doesn't know whether the ownership still belongs to those references or not (e.g. FillEllipse could, in theory, remember the reference it's given, and try to use it at some later moment - remember that language compiler doesn't have any special knowledge of library semantics!).

If you want to indicate that ownership is restricted to that scope, you use the using statement:

using (Bitmap bmp = new Bitmap ( 100, 100 ))
using (Graphics g = Graphics.FromImage ( bmp ))
using (Pen p = new Pen ( Color.FromArgb ( 128, Color.Blue ), 1 ))
using (Brush b = new SolidBrush ( Color.FromArgb ( 128, Color.Blue ) ))
{
    g.FillEllipse ( b, 0, 0, 99, 99 );    
    g.FillRegion ( b, pictureBox1.Region );
}

This will make the compiler insert calls to Dispose automatically as needed, ensuring that all objects are disposed once the corresponding using scope is left (whether normally, by control transfer such as return or break, or an exception).

If you come from a C++ background, using in C# is directly analogous to a const std::auto_ptr, except that it's a language construct here, and can only be used for local variables (i.e. not for class fields).