Hi, On June 13, 2019 11:08:15 AM PDT, didier <did...@gmail.com> wrote: >Extensions can do it, timescaledb in this case with: >INSERT INTO ... RETURNING *; > >Or replacing the test in llvm_compile_expr with an Assert in >slot_compile_deform ?
In that case we ought to never generate a deform expression step - core code doesn't afair. That's only done I'd there's actually something to deform. I'm fine with adding an assert tough Andres -- Sent from my Android device with K-9 Mail. Please excuse my brevity.
