Set/WindowLong problem in user/win test

[Alexandre Julliard]

Francois Gouget <fgouget at free.fr> writes:

> This test uses {Get,Set}WindowLongW which of course does not work on
> Win9x, thus causing the test to fail. I guess part of the goal is to
> check that mixing Ansi calls (CreatewindowExA) with Unicode calls works.

No, there's no real reason to use the W call since we only set parent
and style. We might as well replace all calls by the A version.

> So I wrote a function test_set_window_long function that calls
> SetWindowLongW and then SetWindowLongA and checks that their results
> match but that causes the test to fail (at least on NT4 and Wine).
> More precisely even if all this function does is the following, then the
> test fails:
> 
> {
>     LONG rc=SetWindowLongA(hwnd,nIndex,dwNewLong);
>     SetWindowLongA(hwnd,nIndex,dwNewLong);
>     return rc;
> }
> 
> Comment out the second SetWindowLongA and it works. Why would calling
> SetWindowLong a second time make any difference?

The behavior of SetWindowLong(GWL_HWNDPARENT) depends on the current
parent, so calling it a second time with the same arguments is not
necessarily a nop.

-- 
Alexandre Julliard
julliard at winehq.com