[PATCH 0/4] MR233: win32u: Partially move raw input APIs.

[Rémi Bernon (@rbernon)]

Rémi Bernon (@rbernon) commented about dlls/user32/tests/input.c:
> +                ok(count == ~0u, "GetRawInputData returned %d\n", count);
> +                ok(GetLastError() == ERROR_INVALID_PARAMETER, "GetRawInputData returned %08lx\n", GetLastError());
> +            }
> +            else if (is_wow64)
> +            {
> +                count = GetRawInputData((HRAWINPUT)lparam, RID_INPUT, &ri, &size, sizeof(RAWINPUTHEADER64));
> +                todo_wine ok(count == sizeof(ri), "GetRawInputData returned %d\n", count);
> +                ok(ri.data.mouse.lLastX == 6, "Unexpected rawinput data: %ld\n", ri.data.mouse.lLastX);
> +                todo_wine ok(GetLastError() == 0xdeadbeef, "GetRawInputData returned %08lx\n", GetLastError());
> +            }
> +            else
> +            {
> +                count = GetRawInputData((HRAWINPUT)lparam, RID_INPUT, &ri, &size, sizeof(RAWINPUTHEADER64));
> +                ok(count == ~0u, "GetRawInputData returned %d\n", count);
> +                ok(GetLastError() == ERROR_INVALID_PARAMETER, "GetRawInputData returned %08lx\n", GetLastError());
> +            }
It feels a bit unbalanced to test invalid cases at the same time as a valid case depending on the arch. Could we have instead a test for the invalid sizes, and the another one with valid size?

IMHO using a ternary operator and a variable for the header size instead of two different `GetRawInputData` calls would make it simpler too.

-- 
https://gitlab.winehq.org/wine/wine/-/merge_requests/233#note_1936