Re: Performance degradation in commit ac1d794

On 2016-03-17 09:01:36 -0400, Robert Haas wrote:
> > * Right now the caller has to allocate the WaitEvents he's waiting for
> > locally (likely on the stack), but we also could allocate them as part
> > of the WaitEventSet. Not sure if that'd be a benefit.
>
> I'm not seeing this. What do you mean?

Right now, do use a WaitEventSet you'd do something like
WaitEvent event;

ModifyWaitEvent(FeBeWaitSet, 0, waitfor, NULL);

WaitEventSetWait(FeBeWaitSet, 0 /* no timeout */, &event, 1);

i.e. use a WaitEvent on the stack to receive the changes. If you wanted
to get more changes than just one, you could end up allocating a fair
bit of stack space.

We could instead allocate the returned events as part of the event set,
and return them. Either by returning a NULL terminated array, or by
continuing to return the number of events as now, and additionally
return the event data structure via a pointer.

So the above would be

WaitEvent *events;
int nevents;

ModifyWaitEvent(FeBeWaitSet, 0, waitfor, NULL);

nevents = WaitEventSetWait(FeBeWaitSet, 0 /* no timeout */, events, 10);

for (int off = 0; off <= nevents; nevents++)
; // stuff

Andres