Re: bonobo-activation bug ...

[Michael Meeks <michael peabody ximian com>]

Hi Maciej,

	Ok - so it seems to fix this at all properly is extremely
non-trivial, and a hack around it is utterly evil; to replicate the
problem simply:

	g_idle_add (some_handler_that_does_bonobo_activation_work);
	bonobo_activation_do_some_query ();

	And watch 2+ servers arrive; and confuse everything.

	I can fix this - but it will result in a large scale re-work of
bonobo-activation; and I'm not prepared to do this;

	Regards,

		Michael.

-- 
 mmeeks gnu org  <><, Pseudo Engineer, itinerant idiot

On Thu, 29 Nov 2001, Michael Meeks wrote:

>
> Hi Maciej,
>
> 	We're still having issues with activating multiple
> bonobo-activations if people do work that uses bonobo-activation in an
> idle handler while the server is started.
>
> 	It's really not quite clear how to deal with this race
> properly - I think we need an internal re-work and make the
> registering of results from activated servers done via. callbacks.  We
> also need some sort of batching to ensure that IOR's don't get sent
> back to the activating server until a server process has registered
> all of it's servers.
>
> 	Either way, I'll try and hack around the local case - since
> it's only really a problem for bonobo-activation - since the
> activators code is rather aesoteric.
>
> 	Sigh,
>
> 		Michael.
>
> --
>  mmeeks gnu org  <><, Pseudo Engineer, itinerant idiot
>
>