Bob Jones wrote: > My current thinking is that the filter is a bit like an "fgrep" and > the sort actually requires memory allocation and some "real work", and > thus I've settled on the filter for now pending experiments with a > larger quantity of data.
That's fine. A top-1-sort is less work than you maybe think: You go through all items and find the biggest one. So there is one comparison operator per row - very similar to what happens when "grepping" for NULL values. Yours, Laurenz Albe -- Cybertec | https://www.cybertec-postgresql.com