Re: Fixing the Gst::Message classes

[José Alburquerque <jaalburquerque cox net>]

Murray Cumming wrote:
> On Wed, 2007-12-12 at 13:34 -0500, José Alburquerque wrote:   
>   
> > > I'm sorry, Murray.  What's happening is that the old version uses 
> > > message->get_type() to get the message type when it should be using 
> > > message->get_message_type().  Sorry about that.
> > >
> > > -Jose
> > >   
> > >       
> > And I can confirm that it does work when the call to message->get_type() 
> > is changed.
> >     
>
> Now I am even more confused. Is there a patch or a new test that I
> should commit.
>
>   
No, I think that the slot version of the player is fine.  I think that 
what happened is that when we re-wrapped Gst::Message to use the new 
Gst::MiniObject wrapping functionality, one of us changed the 
_MEMBER_GET(type, ...) to _MEMBER_GET(message_type, ...) renaming the 
old "Gst::MessageType get_type()" method to "Gst::MessageType 
get_message_type()".  Also, Gst::Message has a "GType get_type()" method 
(which I think comes from Gst::MiniObject) which returns a valid integer 
(but not a Gst::MessageType).  These changes broke the old copy of the 
signal version of the player because the old copy thinks that 
"Gst::Message::get_type()" returns a "Gst::MessageType" when in fact it 
now returns a "GType" (the on_bus_message() slot never got an error or 
EOS message so it would never quit the mainloop).

When I made the correct change in the old copy signal player (from 
messsage->get_type() to message->get_message_type()) things worked fine 
so there is no error and no need to change anything.  I hope this all 
makes sense. :-)

-Jose