Re: Bad planner performance for tables with empty tuples when using JIT

I would do this, until PG14 and you can verify it works in most cases for you. 9.6 to 13 is already a big jump and you are already benefiting from many other changes. No need to opt-in to one that you know causes pain (some of the time). Else, try increasing jit_above_cost significantly.

Given this is a feature (and also touches quite a few files), it would not be backpatched afaik.