By adding some debug info in cally, I also noticed that Cally really respond the pyatspi command and get right result on App side. But from dbus-monitor info, seems the result is not passed to test script side at all. So, the culprit is Pyatspi Cache. My question is, your fix will go to pyatspi or Cally. Thanks
WangJing -----Original Message----- From: Piñeiro [mailto:apinhe...@igalia.com] Sent: 2009年10月15日 19:09 To: Wang, Jing J Cc: Wan, Shuang; clutter@o-hand.com Subject: Re: [clutter] Re: CALLY over ATSPI-dbus From: "Wang, Jing J" <jing.j.w...@intel.com> > Piñeiro, > Another problem I met is, with atspi-dbus, the example > cally-atkcomponent-example only expose Application and Main Stage to outside, > 3 rectangle button actors can't be detected. > I found if I moved cally_util_a11y_init from the beginning to end of > the source file (before clutter_main()). The 3 rectangle button actors can be > detected successfully from pyatspi. > Seems Cally can't realize the actors which are added after Cally Module > is loaded. Is that what you mentioned "add_global_event_listener" issue? If > yes, hope it can be fixed ASAP. Yes, is about the listeners, please take a look to the whole thread opened by Shuang Wang: http://lists.o-hand.com/clutter/3241.html A summary: The problem here is *not* that Cally can't realize about the actors added. If you ask directly to the Cally objects you will have the correct values, and in fact, it emit the events about being added. The problem is that the listener is not supported, and as pytatspi makes a cache of the current hierarchy tree (something really sensible), updated with the event listener, the tree is not properly updated because the missing event listener. And yes, this is a bug that I really want to solve. BTW: Mark Doffman commented that he is thinking in adding a non-caching mode, that would solve the problem as well. BR > Thanks > > WangJing > > -----Original Message----- > From: Piñeiro [mailto:apinhe...@igalia.com] > Sent: 2009年10月13日 22:14 > To: Wang, Jing J > Cc: Wan, Shuang; clutter@o-hand.com > Subject: Re: [clutter] Re: CALLY over ATSPI-dbus > > From: "Wang, Jing J" <jing.j.w...@intel.com> > > > Piñeiro, > > Thanks for your help. I just tried the example --with-dbus, and it > > works fine. > > Next, another problem which may block Clutter GUI Auto Testing is How > > to load Cally and atk adaptor module without apps modification, just as the > > case of gtk application by --gtk-module to load external module. > > Yes, this is a know issue, check this bugs [1][2]. > > One option, although somewhat hackish, is keep using GTK_MODULES, and > just keep installing the cally module in the gtk directory (as says > [1]), and keep using GTK_MODULES. In some environments I used that, > but it is clear that it is a hack-short-term solution. Something like: > > GTK_MODULES=cally:atk-bridge (with bonobo) > GTK_MODULES=cally:spiatk (with DBUS) > > Although for the last one, I still need to fix this > add_global_event_listener problem. > > > I heard you plan to integrate cally into Clutter, really? > > Hmm, I think that I don't understand you. Right now Cally is in the > o-hand repository, like other clutter add-on (clutter-gst, etc). > > If you are talking about the run-time, Cally is as integrated with > Clutter as GAIL is integrated with Gtk+. This have the advantage that > you can load the a11y support only if you require that. > > You mean that you would like a deeper integration? > > BR > > [1] http://bugzilla.o-hand.com/show_bug.cgi?id=1737 > [2] http://bugzilla.o-hand.com/show_bug.cgi?id=1738 > > === > API (apinhe...@igalia.com)
