This updated version works for me and addresses previous comments.
I think that such tests are definitely valuable, especially as many corner cases which must trigger errors are covered.
I recommend to apply it.
Please find an updated patch as per comments on Commitfest (comments replicated below for ease of understanding). Feedback 1: fc: role_ro2/3 used twice? rt: Corrected in this update. Feedback 2: fc: I do not understand why "asdf" conveys anything about an expected failure. Association of Scientists, Developers and Faculties? :-) rt: ASDF is a pattern that I learnt in one of the tests (SEQUENCE?) that pre-existed when I started working. Its a slang for arbit text that I just reused thinking that it is normal practice here. Anyway, have corrected that in this update. Feedback 3: fc: 2030/1/1 -> 2030-01-01? maybe use a larger date? rt: 2030/1/1 date is not a failure point of the test. It needs to be a valid date (but sufficiently distant that so that tests don't fail). I tried setting this to 2200/1/1 and I get the same error message. Let me know if this still needs to be a large date. fb: VALID UNTIL '9999-12-31' works for me... rt: I thought 20 years is a date sufficiently far ahead to ensure that this test doesn't fail. Sure, have updated the test to use 9999/1/1. Also, have added more tests at the end to ensure date-checks are also being validated in ALTER ROLE VALID UNTIL.
-- Fabien. -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers