Re: [g-a-devel]Gail menu Patch

["Padraig O'Briain" <Padraig Obriain sun com>]

Marc,

Thanks for this.

Please commit.

Padraig

> Hi All,
> 
> This Gail patch fixes a few problems I've discovered with menus.  It makes 
> GailItem use gtk_label_get_text for the accessible name-- this gets rid of 
> underscores showing up in Menu item names.
> 
> When creating a GailMenuItem, this patch sets the role of the AtkObject to 
> ATK_ROLE_MENU when the corresponding GtkMenuItem has a submenu.
> 
> A more serious problem-- because menu items don't receive "real" GTK+ 
> focus, Gail traps the select signal on menu items.  The undesireable side 
> affect of this is that when the open menus disappear, the AT client is not 
> notified that the keyboard focus has returned to the widget which had it 
> prior to the menus being posted.  This problem can be illustrated by 
> pressing f10 to post the menus, arrowing arround in the menus, and then 
> pressing esc. to return to the widget having the focus prior to pressing 
> f10.  This patch attempts to fix this as follows:
> 
> 1. When the widget receiving the "select" signal is a GtkMenuItem, and the 
> widget known to currently have the "ATK" focus is not a GtkMenuItem, then I 
> cache a copy of the item known to have focus.
> 2. I added a handler for "deactivate" signals on GtkmenuShell widgets.  If 
> the GtkMenuShell receiving the deactivate signal has no parent menu shell, 
> meaning that it is the top level menubar, I fake an atk focus signal to the 
> AtkObject representing the widget which I cached.
> 
> The only side affect I can see is that when a menu item is activated, and 
> for example, brings up a dialog box, two ATK focus signals will be 
> emitted-- one bringing ATK focus back to the widget having it before the 
> menus were posted, and the second to the dialog.  IMHO, this is much better 
> than an AT getting lost when the menus disappear.
> 
> Marc