Yes! Exactly my point.
-> richard
Todor Boev wrote:
Yup. It's just an "attempt" to make unregistration work - a failed
one. Lately I am leaning ever more to the view that the only
reasonable way to use services is to do it opportunistically like so:
http://code.google.com/p/peaberry/issues/detail?id=27&can=1#c3
In this light all the hubbub with the synchronous events goes away.
Now the user needs
1) a "registered" event to start trying to use the service.
2) an "unregistered" event to stop trying to use the service and
conserve CPU resources.
Here we only need to guarantee that "registered" always precedes
"unregistered". Too bad the super-strong synchronous design is already
in the spec. It's always easier to strengthen the constraints than to
weaken them.
Cheers,
Todor
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
