strk,
about the failure in key_event_test.c:
found the culprint. It is the ActionExec::cleanupAfterRun(bool
expectInconsistencies). See the following call stack:
...
movie_root::processActionQueue()
movie_root::flushHigherPriorityActionQueues()
ActionExec::cleanupAfterRun()
ActionExec::operator()()
swf_function::operator()
call_method()
seems ActionExec::cleanupAfterRun() does too much as its name
suggests. Why cleanupAfterRun triggers processing the action queue?
I guess this is not only specific to key event handlers, other
function calls might have the same problem. eg. call_method() triggers
processing actions queue, which is not the expected behaviour. Can
you improve the document of this function?
--zou
On Nov 22, 2007 3:29 PM, strk <[EMAIL PROTECTED]> wrote:
> On Thu, Nov 22, 2007 at 01:35:16PM +0800, zou lunkai wrote:
> > > I guess all event handlers are *pushed* instead of executed
> > > immediatly in cvs-head.
> >
> > Not true. But confirmed key_event_test was failed due to actions order.
>
> I still get failures with the pp sometimes.
> The test should probably be more predictable.
>
> --strk;
>
_______________________________________________
Gnash-commit mailing list
Gnash-commit@gnu.org
http://lists.gnu.org/mailman/listinfo/gnash-commit
