Hello, Dave Cridland wrote:
> Could you elaborate on that? Our PEP implementation is certainly not > a full pubsub implementation, so it's entirely possible I've missed > what you're hinting at here, but nothing springs to mind. > > (Feel free to raise it on standards@, it's probably more suitable > there.) In real life we have seen that misbehaving client can generate huge number of pending queries. It means that you have to keep track of them until you time them out. Other problems thtt comes to mind happens when you put s2s into the mix. You cannot always be sure you get disconnect event so you have problem with the caps clean up strategy. -- Mickaël Rémond http://www.process-one.net/
