Greg Stark <gsstark@mit.edu> writes:
> Are we still going to subtract out the gettimeofday overhead?
> I was always more excited about that than the sampling aspect. I've run into
> queries where EXPLAIN ANALYZE results were deceptive due to the gettimeofday
> overhead but I've never run into a query where gettimeofday overhead made
> running the query under EXPLAIN ANALYZE impractical.
That would be deceptive in a different way, ie, make it look like there
was a whole bunch of outside-the-plan overhead.
regards, tom lane