Re: supporting notifications from backends

[Cyrille Moureaux <Cyrille Moureaux Sun COM>]

Hi Mark,

> 	Ah, I hadn't thought of that extra complexity for the backends. Because
> of that and the fact that, as you say, allowing a different callback per
> listener is useless I've added back a set_notify_func() vtable member.
>  
I think we also need to mention in the documentation that the 
notification function will only be called once per change relevant to at 
least one listener, and not once per registered listener for that 
change. In other words, if there are two listeners for 
/apps/metacity/general, one for /apps/metacity and twenty lazy ones for 
/apps, the notification function should be called once by the backend if 
something changes in /apps/metacity/general and not twenty-three times.

> 	Attaching how she's looking now. Let me know if you see any further
> problems with this. 
>  
It seems the set_notify_func member of the vtable is not in the 
documentation, and the precision I've mentioned above should probably be 
along your note that changes made by the daemon itself should not 
trigger a notification.

> 	It should be in CVS by tommorrow. I'd really appreciate it if you could
> try updating your backend then and making sure everything is okay.
>
>  
Will do. Thanks,

Cyrille