* Alvaro Herrera (alvherre@2ndquadrant.com) wrote:
> Tom Lane wrote:
> > In general, we might want to consider replacing long sleep intervals
> > with WaitLatch operations. I thought for a bit about trying to turn
> > pg_usleep itself into a WaitLatch call; but it's also used in frontend
> > code where that wouldn't work, and anyway it's not clear this would be
> > a good thing for short sleeps.
>
> How about having a #ifdef !FRONTEND code path that uses the latch, and
> sleep otherwise? And maybe use plain sleep for short sleeps in the
> backend also, to avoid the latch overhead. I notice we already have
> three implementations of pg_usleep.
Is there really serious overhead from using latches..? I thought much
of the point of that approach was specifically to minimize overhead...
Thanks, Stephen