Created attachment 194628 [details] [review]
Using the proper format to register the events

This patch is conservative, as it also tries to register to the window events with the old format. I think that this is required until all ATK implementors moves to the new format.

In order to do that I just try to register using the old format. If it was a success we assume that the ATK implementor is "old". This should work, although the bad news is that you would get a warning with an updated ATK implementor.

Comment on attachment 194628 [details] [review]
Using the proper format to register the events






>From cb2f09781d865393ed7866f9499fa65ce819a04d Mon Sep 17 00:00:00 2001
>From: =?UTF-8?q?Alejandro=20Pi=C3=B1eiro?= <apinheiro@igalia.com>
>Date: Tue, 23 Aug 2011 17:07:02 +0200
>Subject: [PATCH] Using new format to register to window events
>
>Now ATK has AtkWindow, so the format to register to window events
>is like any other ATK event
>
>We check if we need to register using the old format, assuming that
>if you are able to register to a window event using the old format
>is because the ATK implementor is still using the old one. This
>check should be temporal.
>---
> atk-adaptor/event.c |   48 +++++++++++++++++++++++++++++++++++++++---------
> configure.ac        |    2 +-
> 2 files changed, 40 insertions(+), 10 deletions(-)
>
>diff --git a/atk-adaptor/event.c b/atk-adaptor/event.c
>index f5a2d2d..af1943b 100644
>--- a/atk-adaptor/event.c
>+++ b/atk-adaptor/event.c
>@@ -2,6 +2,7 @@
>  * AT-SPI - Assistive Technology Service Provider Interface
>  * (Gnome Accessibility Project; http://developer.gnome.org/projects/gap)
>  *
>+ * Copyright 2011, F123 Consulting & Mais Diferenças
>  * Copyright 2008, 2009, Codethink Ltd.
>  * Copyright 2001, 2002, 2003 Sun Microsystems Inc.,
>  * Copyright 2001, 2002, 2003 Ximian, Inc.
>@@ -1068,13 +1069,17 @@ generic_event_listener (GSignalInvocationHint * signal_hint,
>  * and stores the signal id returned so that the function may be
>  * de-registered later.
>  */
>-static void
>+static guint
> add_signal_listener (GSignalEmissionHook listener, const char *signal_name)
> {
>   guint id;
> 
>   id = atk_add_global_event_listener (listener, signal_name);
>-  g_array_append_val (listener_ids, id);
>+
>+  if (id > 0) /* id == 0 is a failure */
>+    g_array_append_val (listener_ids, id);
>+
>+  return id;
> }
> 
> /*
>@@ -1091,6 +1096,7 @@ spi_atk_register_event_listeners (void)
>    */
>   GObject *ao = g_object_new (ATK_TYPE_OBJECT, NULL);
>   AtkObject *bo = atk_no_op_object_new (ao);
>+  guint id = 0;
> 
>   g_object_unref (G_OBJECT (bo));
>   g_object_unref (ao);
>@@ -1108,13 +1114,37 @@ spi_atk_register_event_listeners (void)
> 
>   add_signal_listener (property_event_listener,
>                        "Gtk:AtkObject:property-change");
>-  add_signal_listener (window_event_listener, "window:create");
>-  add_signal_listener (window_event_listener, "window:destroy");
>-  add_signal_listener (window_event_listener, "window:minimize");
>-  add_signal_listener (window_event_listener, "window:maximize");
>-  add_signal_listener (window_event_listener, "window:restore");
>-  add_signal_listener (window_event_listener, "window:activate");
>-  add_signal_listener (window_event_listener, "window:deactivate");
>+
>+  /* window events: we tentative try to register using the old format */
>+  id = add_signal_listener (window_event_listener, "window:create");
>+
>+  if (id != 0)
>+    {
>+      /* If we are able to register using the old format, we assume
>+       * that the ATK implementor is managing window events without
>+       * AtkWindow. We can't use the opposite test because after
>+       * including AtkWindow on ATK you would be able to register to
>+       * that event, although the ATK implementor could or not use it.
>+       */
>+
>+      add_signal_listener (window_event_listener, "window:destroy");
>+      add_signal_listener (window_event_listener, "window:minimize");
>+      add_signal_listener (window_event_listener, "window:maximize");
>+      add_signal_listener (window_event_listener, "window:restore");
>+      add_signal_listener (window_event_listener, "window:activate");
>+      add_signal_listener (window_event_listener, "window:deactivate");
>+    }
>+  else
>+    {
>+      add_signal_listener (window_event_listener, "Atk:AtkWindow:create");
>+      add_signal_listener (window_event_listener, "Atk:AtkWindow:destroy");
>+      add_signal_listener (window_event_listener, "Atk:AtkWindow:minimize");
>+      add_signal_listener (window_event_listener, "Atk:AtkWindow:maximize");
>+      add_signal_listener (window_event_listener, "Atk:AtkWindow:restore");
>+      add_signal_listener (window_event_listener, "Atk:AtkWindow:activate");
>+      add_signal_listener (window_event_listener, "Atk:AtkWindow:deactivate");
>+    }
>+
>   add_signal_listener (document_event_listener,
>                        "Gtk:AtkDocument:load-complete");
>   add_signal_listener (document_event_listener, "Gtk:AtkDocument:reload");
>diff --git a/configure.ac b/configure.ac
>index 096e98e..91c2497 100644
>--- a/configure.ac
>+++ b/configure.ac
>@@ -51,7 +51,7 @@ PKG_CHECK_MODULES(GMODULE, [gmodule-2.0 >= 2.0.0])
> AC_SUBST(GMODULE_LIBS)
> AC_SUBST(GMODULE_CFLAGS)
> 
>-PKG_CHECK_MODULES(ATK, [atk >= 2.1.0])
>+PKG_CHECK_MODULES(ATK, [atk >= 2.1.5])
> AC_SUBST(ATK_LIBS)
> AC_SUBST(ATK_CFLAGS)
> 
>-- 
>1.7.5.4
>

Sorry for the delay--this slipped my mind until now.  The patch seemed okay to me from my brief testing, so I've committed it for the release.