Hello I think the tuplecid.sql test is rather brute-force-ish, and I'm not sure we need it. To verify that I ran the tests with it and without it, and compared the difference in the HTML coverage report. I see no additional extra lines marked as covered with the tuplecid.sql left in place, which I take to mean to support my hypothesis. I suggest we throw it away and keep only the two isolation tests.
-- Álvaro Herrera PostgreSQL Developer — https://www.EnterpriseDB.com/ "Las cosas son buenas o malas segun las hace nuestra opinión" (Lisias)
