Laurenz Albe <laurenz.a...@cybertec.at> writes: > I wonder if it is worth the extra planning time to detect and improve > such queries.
I'm skeptical too. I'm *very* skeptical of implementing it in the grammar as shown here --- I'd go so far as to say that that approach cannot be accepted. That's far too early, and it risks all sorts of problems. An example is that the code as given seems to assume that all the sublists are the same length ... but we haven't checked that yet. I also suspect that this does not behave the same as the original construct for purposes like resolving dissimilar types in the VALUES list. (In an ideal world, perhaps it'd behave the same, but that ship sailed a couple decades ago.) regards, tom lane