bug#23568: 25.0.94; Mode line menus appear incorrectly in some monitor configurations

[Alex]

Alex <agrambot@gmail.com> writes:

> This is still an issue in current master.
>
> The tooltip issue I mentioned above last year was fixed in 7b14da444, so
> I tried using pretty much the same code to fix this issue; a proof of
> concept diff is included below.
>
> I don't know if it's a good approach though. Partially because I have no
> idea why the values of req.{height, width} are so large, why they differ
> in different Emacs sessions, and how they relate to the other
> width/height variables.
>
> In any case, perhaps this functionality should be abstracted into a
> procedure that just computes the dimensions of the current monitor?
>
> diff --git a/src/xmenu.c b/src/xmenu.c
> index 2805249164..9a87fce380 100644
> --- a/src/xmenu.c
> +++ b/src/xmenu.c
> @@ -1161,8 +1161,43 @@ menu_position_func (GtkMenu *menu, gint *x, gint *y, 
> gboolean *push_in, gpointer
>    struct next_popup_x_y *data = user_data;
>    GtkRequisition req;
>    struct x_display_info *dpyinfo = FRAME_DISPLAY_INFO (data->f);
> -  int disp_width = x_display_pixel_width (dpyinfo);
> -  int disp_height = x_display_pixel_height (dpyinfo);
> +  int disp_width;
> +  int disp_height;
> +
> +  int min_x, min_y, max_x, max_y = -1;
> +  Lisp_Object frame, attributes, monitor, geometry;
> +
> +  XSETFRAME(frame, data->f);
> +  attributes = Fx_display_monitor_attributes_list (frame);
> +
> +  /* Try to determine the current monitor's resolution */
> +  while (CONSP (attributes))
> +    {
> +      monitor = XCAR (attributes);
> +      geometry = Fassq (Qgeometry, monitor);
> +      if (CONSP (geometry))
> +        {
> +          min_x = XINT (Fnth (make_number (1), geometry));
> +          min_y = XINT (Fnth (make_number (2), geometry));
> +          max_x = min_x + XINT (Fnth (make_number (3), geometry));
> +          max_y = min_y + XINT (Fnth (make_number (4), geometry));
> +          if (min_x <= data->x && data->x < max_x
> +              && min_y <= data->y && data->y < max_y)
> +            {
> +              disp_width = max_x;
> +              disp_height = max_y;
> +              break;
> +            }
> +          max_y = -1;
> +        }
> +
> +      attributes = XCDR (attributes);
> +    }
> +  if ( max_y < 0 )
> +    {
> +      disp_width = x_display_pixel_width (dpyinfo);
> +      disp_height = x_display_pixel_height (dpyinfo);
> +    }
>  
>    *x = data->x;
>    *y = data->y;

Sorry, I copied the wrong commit hash. I meant to write that it was
fixed in c77ffc8019, which fixed bug#22549.

Martin, since you were involved with bug#22549, what do you think? Does
this seem reasonable, and should this functionality (window dimensions
of the current position's monitor) be extracted into its own procedure?