On Qui, 2007-09-06 at 00:41 +0200, Sebastian Pölsterl wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Hi!
>
> I'm working on Deskbar-Appplet and ran into some problems using the
> beagle handler. I discussed it thoroughly with the beagle maintainers
> and it doesn't seem to be their fault, because the same code worked before.
>
> What happens is:
> I query beagle async like that:
> <code>
> def query(self, qstring):
> beagle_query = beagle.Query()
> beagle_query.add_text(qstring)
> beagle_query.connect("hits-added", self.hits_added_cb, qstring)
> self.beagle.send_request_async(beagle_query)
> </code>
> So the beagle_query variable is local.
>
> The problem is that way I don't receive any results. Further
> investigation revealed that the connection from beagle to deskbar is
> closed before the results can be transmitted.
>
> If I use self.beagle_query instead of beagle_query it works fine.
>
> My assumption is that the garbage collector collects beagle_query at the
> end of the method. If beagle wants to send its results it can't find the
> object anymore, because it has been collected.
> Whereas with self.beagle_query the object doesn't get collected and
> everything works fine.
I also think your assumption is right. beagle_query, as a free
variable, is unref'ed by python as soon as the method goes out of scope,
while self.beagle_query remains referenced from the class instance and
so the GObject wrapped by self.beagle_query is not unreffed.
Why did it work in PyGObject < 2.11? Probably because before wrapped
GObjects did not _usually_ get immediately collected by python, and so
the likely reference counting bug towards the BeagleQuery GObject was
not noticed. Although I believe that even before there was still a
small change that the query GObject would still be collected before
time.
If my theory is correct then this only means that PyGObject 2.11 only
uncovered a bug that also existed before, but which out of luck hardly
ever manifested itself.
>
> Please be aware that this is just an assumption, but if I'm right _and_
> you can fix that bug I would be enormously happy.
I really suspect that there is something wrong with the beagle bindings.
beagle_send_request_async should probably be g_object_ref'ing the
beagle_query object, and unref it later after the response is received
and after calling the async handler.
I hope this helps.
--
Gustavo J. A. M. Carneiro
<[EMAIL PROTECTED]> <[EMAIL PROTECTED]>
"The universe is always one step beyond logic" -- Frank Herbert
_______________________________________________
pygtk mailing list pygtk@daa.com.au
http://www.daa.com.au/mailman/listinfo/pygtk
Read the PyGTK FAQ: http://www.async.com.br/faq/pygtk/
