On Mon, Oct 12, 2020 at 6:35 PM Greg Nancarrow <[email protected]> wrote: > On Mon, Oct 12, 2020 at 2:11 PM Thomas Munro <[email protected]> wrote: > > Yeah, I think we should go ahead and fix that up front. Here's a > > version with a commit message. > > I've checked it and tested it, and it looks fine to me. > Also, it seems to align with the gripe in the old comment about width > ("XXX this is totally wrong: we should report zero if no RETURNING > ..."). > I'm happy for you to commit it.
Pushed, though I left most of that comment there because the width estimate still needs work when you do use RETURNING. At least we now have rows=0 for queries without RETURNING, which was a bigger problem for your patch.
