Re: [PATCH v2 0/5] MR4092: user32: Actually implement RealGetWindowClass{A/W}.

[Connor McAdams (＠cmcadams)]

On Mon Oct 16 13:30:46 2023 +0000, Jinoh Kang wrote:
> In short,
> test_real_class_name() {
>     for (i = 0; i < 2; i++) {
>         pass_msg_to_real_class_wndproc = i;
>         if (!i)
>             test_hwnd_real_class_name_a_w(hwnd, SUPER_CLASS_NAME_W, SUPER_CLASS_NAME_A);
>         else
>             test_hwnd_real_class_name_a_w(hwnd,
> class_test->real_class_name_w, class_test->real_class_name_a);
>     }
> }
> test_real_class_names() {
>     for (i = 0; i < tests_count; i++)
>         test_real_class_name();
> }
>
> This should be just
> test_real_class_name() {
>     for (i = 0; i < 2; i++) {
>         test_hwnd_real_class_name_a_w(hwnd,
> class_test->real_class_name_w, class_test->real_class_name_a);
>     }
> }
> test_real_class_names() {
>     pass_msg_to_real_class_wndproc = 0;
>     test_hwnd_real_class_name_a_w(hwnd, SUPER_CLASS_NAME_W, SUPER_CLASS_NAME_A);
>     pass_msg_to_real_class_wndproc = 1;
>     for (i = 0; i < tests_count; i++) {
>         test_real_class_name();
>     }
> }
>
> Note that the `pass_msg_to_real_class_wndproc = 0` case is being tested
> only once in the latter code snippet.

Ah, yeah that makes sense. I guess I was mainly going for a clear before->after case here, but doing the first case on each class is definitely redundant.