Re: [evolution-patches] gal, accessibility,focused objects should have a state of ATK_STATE_FOCUSED,55249

[JP Rosevear <jpr ximian com>]

On Mon, 2004-03-29 at 07:29, Yuedong Du wrote:
> Hi Mike and all,
> 
> The bug cause gnopernicus cannot read mail in gnome2.6. This is because 
> in gnopernicus > 0.71,
> the focus object must have state FOCUS, or it will not be read.
> 
> So the fix is just to add the state. Also add some sanity check code,

> 
> ______________________________________________________________________
> Index: gal-a11y-e-cell-text.c
> ===================================================================
> RCS file: /cvs/gnome/gal/gal/a11y/e-table/gal-a11y-e-cell-text.c,v
> retrieving revision 1.4
> diff -u -r1.4 gal-a11y-e-cell-text.c
> --- gal-a11y-e-cell-text.c	2 Dec 2003 07:20:04 -0000	1.4
> +++ gal-a11y-e-cell-text.c	29 Mar 2004 12:24:56 -0000
> @@ -107,9 +107,12 @@
>  ect_get_caret_offset (AtkText *text)
>  {
>  	GalA11yECell *gaec = GAL_A11Y_E_CELL (text);
> -	ECellText *ect = E_CELL_TEXT (gaec->cell_view->ecell);
> +	ECellText *ect = NULL;
>  	gint start, end;
>  
> +	g_return_val_if_fail (gaec && gaec->cell_view && gaec->cell_view->ecell && E_IS_CELL_TEXT (gaec->cell_view->ecell), -1);
> +	ect = E_CELL_TEXT (gaec->cell_view->ecell);
> +
>  	if (e_cell_text_get_selection (gaec->cell_view,
>  				       gaec->view_col, gaec->row,
>  				       &start, &end)) {
> Index: gal-a11y-e-table-item.c
> ===================================================================
> RCS file: /cvs/gnome/gal/gal/a11y/e-table/gal-a11y-e-table-item.c,v
> retrieving revision 1.9
> diff -u -r1.9 gal-a11y-e-table-item.c
> --- gal-a11y-e-table-item.c	2 Dec 2003 08:25:17 -0000	1.9
> +++ gal-a11y-e-table-item.c	29 Mar 2004 12:24:58 -0000
> @@ -1035,8 +1035,10 @@
>  			   int index_in_parent)
>  {
>  	GalA11yETableItem *a11y;
> +	AtkObject *accessible;
>  	int n;
>  
> +	g_return_val_if_fail (item && item->cols >= 0 && item->rows >= 0, NULL);
>  	a11y = g_object_new (gal_a11y_e_table_item_get_type (), NULL);
>  
>  	atk_object_initialize (ATK_OBJECT (a11y), item);
> @@ -1044,7 +1046,9 @@
>  	GET_PRIVATE (a11y)->parent = parent;
>  	GET_PRIVATE (a11y)->index_in_parent = index_in_parent;
>  
> -	g_return_val_if_fail (item->cols >= 0 && item->rows >= 0, NULL);
> +
> +	accessible  = ATK_OBJECT(a11y);
> +	accessible->role = ATK_ROLE_TREE_TABLE;
>  	/* Initialize cell data. */
>  	n = item->cols * item->rows;
>  	GET_PRIVATE (a11y)->cols = item->cols;
> @@ -1187,10 +1191,12 @@
>  	g_signal_emit_by_name (a11y, "selection_changed");
>  
>          cell = atk_table_ref_at (ATK_TABLE (a11y), row, col);
> -        if (ATK_IS_OBJECT (cell))
> -                g_signal_emit_by_name  (a11y,
> -                                        "active-descendant-changed",
> -                                        cell);
> +	if (cell != NULL) {
> +        	if (ATK_IS_OBJECT (cell)) {
> +			gal_a11y_e_cell_add_state(cell, ATK_STATE_FOCUSED, FALSE);
> +			atk_focus_tracker_notify (cell);
> +		}
> +	}
>  }

Looks basically fine - why would the be NULL here?  and if it could be,
shouldn't the if (cell != NULL) check be prior to the ATK_IS_OBJECT
check?

-JP
-- 
JP Rosevear <jpr ximian com>
Ximian, Inc.