Re: [PATCH] user32: Do not call SendMessage() to show a window that is already visible.

[Dmitry Timoshkov]

Kimmo Myllyvirta <kimmo.myllyvirta(a)gmail.com> wrote:
> diff --git a/dlls/user32/winpos.c b/dlls/user32/winpos.c
> index c2b35ec..64fdfe0 100644
> --- a/dlls/user32/winpos.c
> +++ b/dlls/user32/winpos.c
> @@ -1239,6 +1239,9 @@ BOOL WINAPI ShowWindow( HWND hwnd, INT cmd )
>      if ((cmd == SW_HIDE) && !(GetWindowLongW( hwnd, GWL_STYLE ) & WS_VISIBLE))
>          return FALSE;
>
> +    if ((cmd == SW_SHOW) && (GetWindowLongW( hwnd, GWL_STYLE ) & WS_VISIBLE))
> +        return TRUE;
> +
>      return SendMessageW( hwnd, WM_WINE_SHOWWINDOW, cmd, 0 );
>  }

Please have a look at the show_window() implementation and try to cover
all cases that trigger SWP_SHOWWINDOW flag passed to SetWindowPos() in
order to actually show a window. In fact it's only SW_HIDE that leads
to that, so perhaps just adding an 'else' to the SW_HIDE case is enough?

-- 
Dmitry.

[PATCH] user32: Do not call SendMessage() to show a window that is already visible.

[Kimmo Myllyvirta]

On 04/30/2017 07:38 AM, Dmitry Timoshkov wrote:
> Kimmo Myllyvirta <kimmo.myllyvirta(a)gmail.com> wrote:
> > diff --git a/dlls/user32/winpos.c b/dlls/user32/winpos.c
> > index c2b35ec..64fdfe0 100644
> > --- a/dlls/user32/winpos.c
> > +++ b/dlls/user32/winpos.c
> > @@ -1239,6 +1239,9 @@ BOOL WINAPI ShowWindow( HWND hwnd, INT cmd )
> >       if ((cmd == SW_HIDE) && !(GetWindowLongW( hwnd, GWL_STYLE ) & WS_VISIBLE))
> >           return FALSE;
> >
> > +    if ((cmd == SW_SHOW) && (GetWindowLongW( hwnd, GWL_STYLE ) & WS_VISIBLE))
> > +        return TRUE;
> > +
> >       return SendMessageW( hwnd, WM_WINE_SHOWWINDOW, cmd, 0 );
> >   }
> > Please have a look at the show_window() implementation and try to cover
> > all cases that trigger SWP_SHOWWINDOW flag passed to SetWindowPos() in
> > order to actually show a window. In fact it's only SW_HIDE that leads
> > to that, so perhaps just adding an 'else' to the SW_HIDE case is enough?

Yeah, it would be better to fix all the cases at once instead of 
one-by-one. Probably something similar to the attached diff. I'll take a 
look at it later, this is already fixed in staging it seems.