> I will look at this one. It'd be nice to get it in if at all possible,
> because the existing listen/notify infrastructure won't play very nicely
> with HS --- eg, inspecting pg_listener on the slave might yield the
> false impression that some of the slave-side backends had active LISTENs
> because of chance matches of PID.
It'll also serve a major need for integrating PostgreSQL with caching
infrastructures. So it's not just an "insider" feature.
--Josh Berkus