The Is­Window­Unicode function reports whether the current window procedure expects Unicode window messages. What specifically does this mean?

It means that messages like WM_CHAR will report characters as Unicode code units rather than as ANSI code units.

Okay, so what determines whether a window procedure receives Unicode messages?

When a window is created, its initial window procedure comes from the class registration. If the class was registered with Register­Class[Ex]W, then it is a Unicode window procedure; otherwise it was registered with Register­Class[Ex]A and is an ANSI window procedure.

If a window is subclassed with a function pointer, then the new window procedure is Unicode if it was set with Set­Window­Long­PtrW(GWLP_WNDPROC) and ANSI if it was set with Set­Window­Long­PtrA(GWLP_WNDPROC).

As we saw some time ago, Get­Window­Long­Ptr(GWLP_WNDPROC) returns a thunk if you use the ANSI version but the current window procedure is Unicode, or vice versa. Suppose the window procedure is Unicode and you call Get­Window­Long­PtrA(GWLP_WNDPROC). This will return you a thunk, and if you later set that thunk back as the window procedure with Set­Window­Long­PtrA(GWLP_WNDPROC), then the original Unicode window procedure is restored, and the window is a Unicode window again.

Note that calling Set­Window­Long­PtrA(GWLP_WNDPROC) can result in a window procedure that is Unicode.

As a window message travels from one window procedure to another window procedure (via Call­Window­Proc), the character set may flip between Unicode and ANSI. The Is­Window­Unicode function tells you only about the window’s current window procedure.