On 18 Sep 2026, Kirill Reshke wrote:
> One simple fix can be simply importing NOT NULL NOT VALID as a
> nullable column, but this probably would make some people unhappy.
I support the simpler fix. Of the constraints the local planner could
use to infer properties of the remote rows, postgres_fdw imports only
NOT NULL. Importing only validated ones seems a reasonable policy.
This also needs fixing on older branches importing from a PG18+ server.
Could we adapt the query-result test from your second patch for the
first one?
Thank you!
Best regards, Andrey Borodin.