IDispose和Finalize的区别和联系

using System;
using System.ComponentModel;

// The following example demonstrates how to use the
// GC.SuppressFinalize method in a resource class to prevent
// the clean-up code for the object from being called twice.

public class DisposeExample
{
    // A class that implements IDisposable.
    // By implementing IDisposable, you are announcing that
    // instances of this type allocate scarce resources.
    public class MyResource: IDisposable
    {
        // Pointer to an external unmanaged resource.
        private IntPtr handle;
        // Other managed resource this class uses.
        private Component component = new Component();
        // Track whether Dispose has been called.
        private bool disposed = false;

        // The class constructor.
        public MyResource(IntPtr handle)
        {
            this.handle = handle;
        }

        // Implement IDisposable.
        // Do not make this method virtual.
        // A derived class should not be able to override this method.
        public void Dispose()
        {
            Dispose(true);
            // This object will be cleaned up by the Dispose method.
            // Therefore, you should call GC.SupressFinalize to
            // take this object off the finalization queue
            // and prevent finalization code for this object
            // from executing a second time.
            GC.SuppressFinalize(this);//问题2:在“Dispose()”中“GC.SuppressFinalize(this);”是什么意思?
        }

        // Dispose(bool disposing) executes in two distinct scenarios.
        // If disposing equals true, the method has been called directly
        // or indirectly by a user's code. Managed and unmanaged resources
        // can be disposed.
        // If disposing equals false, the method has been called by the
        // runtime from inside the finalizer and you should not reference
        // other objects. Only unmanaged resources can be disposed.
        private void Dispose(bool disposing)
        {
            // Check to see if Dispose has already been called.
            if(!this.disposed)
            {
                // If disposing equals true, dispose all managed
                // and unmanaged resources.
                if(disposing)
                {
                    // Dispose managed resources.
                    component.Dispose();//问题1:也就是为什么GC垃圾回收机制在回收对象的时候只回收或者释放非托管资源,而不回收托管资源?
                }
        
                // Call the appropriate methods to clean up
                // unmanaged resources here.
                // If disposing is false,
                // only the following code is executed.
                CloseHandle(handle);
                handle = IntPtr.Zero;           
            }
            disposed = true;        
        }

        // Use interop to call the method necessary 
        // to clean up the unmanaged resource.
        [System.Runtime.InteropServices.DllImport("Kernel32")]
        private extern static Boolean CloseHandle(IntPtr handle);

        // Use C# destructor syntax for finalization code.
        // This destructor will run only if the Dispose method
        // does not get called.
        // It gives your base class the opportunity to finalize.
        // Do not provide destructors in types derived from this class.
        ~MyResource()     
        {
            // Do not re-create Dispose clean-up code here.
            // Calling Dispose(false) is optimal in terms of
            // readability and maintainability.
            Dispose(false);//问题1:为什么在析构函数中调用的是“Dispose(false);”?
        }
    }

    public static void Main()
    {
        // Insert code here to create
        // and use a MyResource object.
    }
}

产生了两个问题(代码中红色标注区域):

问题1:为什么在析构函数中调用的是“Dispose(false);”,也就是为什么GC垃圾回收机制在回收对
象的时候只回收或者释放非托管资源,而不回收托管资源?
问题2:在“Dispose()”中“GC.SuppressFinalize(this);”是什么意思?

上网搜了写资料,终于了解了这种设计的用意和优点:
参考资料:
http://baike.baidu.com/view/4471636.htm
http://www.360doc.com/content/11/0503/18/6075898_114106056.shtml

解答问题1:
在.NET的对象中实际上有两个用于释放资源的函数:Dispose和Finalize。Finalize的目的是用于释放非托管的资源,而Dispose是用于释放所有资源,包括托管的和非托管的。
在这个模式中,通过一个变量“disposing”来区分是客户调用(true)还是GC调用(false)。
这是因为,Dispose()函数是被其它代码显式调用并要求释放资源的,而Finalize是被GC调用的。
在GC调用的时候MyResource所引用的其它托管对象(component)可能还不需要被销毁,并且即使
要销毁,也会由GC来调用。因此在Finalize中只需要释放非托管资源即可。

解答问题2:
由于在Dispose()中已经释放了托管和非托管的资源,因此在对象被GC回收时再次调用Finalize是
没有必要的,所以在Dispose()中调用GC.SuppressFinalize(this)避免重复调用Finalize。

因此,上面的模式保证了:
1、 Finalize只释放非托管资源;
2、 Dispose释放托管和非托管资源;
3、 重复调用Finalize和Dispose是没有问题的;
4、 Finalize和Dispose共享相同的资源释放策略,因此他们之间也是没有冲突的。
在C#中,这个模式需要显式地实现,其中C#的~MyResource()函数代表了Finalize()。

优点:
1、如果客户没有调用Dispose(),未能及时释放托管和非托管资源,那么在垃圾回收时,还有机会执
行Finalize(),释放非托管资源,但是造成了非托管资源的未及时释放的空闲浪费。
2、如果客户调用了Dispose(),就能及时释放了托管和非托管资源,那么该对象被垃圾回收时,不回
执行Finalize(),提高了非托管资源的使用效率并提升了系统性能。

此外还有Close()方法,此方法一般和Open()方法配合来使用,对于数据库连接,一般可以逆向操作,比如打开->关闭,关闭->打开,而对于文件操作,一般是关闭文件,相当于Dispose(),而且有的用户更愿意使用Close()来释放资源,所以出现了一下这种相当于适配器模式的代码:

 

public void Close()

{

Dispose(();

}

最后还要再说一点,据MSDN上说C#不允许类实现Finalize()方法,所以用析构函数来代替。

posted @ 2012-12-17 10:22  我不是杰克船长  阅读(322)  评论(0编辑  收藏  举报