Using the DeferWindowPos APIs

There's a set of little known APIs that allow you to move multiple windows in one fell swoop, which can reduce flicker in contrast to moving each window individually. Using the APIs is fairly straight forward, but like many such APIs they require you to use multiple steps. Forget a single step and things won't work as advertised.

C++ programmers use a technique known as "resource allocation is acquisition" to ensure that any "finalization" steps are not forgotten. Basically, the constructor sets things up and the destructor insures the finalization occurs. The most widely known use of this technique is the std::auto_ptr class, which insures the pointer is deleted when the auto_ptr leaves scope.

We can use this same technique to insure that the three steps required for DeferWindowPos are followed. We'll create a class to wrap DeferWindowPos. The constructor for the class will call BeginDeferWindowPos and the destructor will call EndDeferWindowPos. Member functions will call DeferWindowPos using the stored HDWP handle required by these APIs. Calls to these methods will delay the window positioning until the destructor is called. So, to move two windows at once is as easy as the following:

{
	CDeferPos dp(2);	// We'll move 2 windows
	dp.MoveWindow(hWnd1, 0, 0, 10, 20, TRUE);
	dp.MoveWindow(hWnd2, 10, 0, 10, 20, TRUE);
}	// When dp goes out of scope both windows are moved at once

Note that the constructor takes a parameter indicating how many windows we think we'll move. This is just a suggestion that will help Windows to optimize the memory allocation. If you don't know how many windows you'll move you can either guess or leave it to the default of 1. Windows will increase the memory size as needed when you move more windows than initially allocated.

For the complete code listing for CDeferPos check out the download.