Michael Paquier <michael@paquier.xyz> writes:
> On Thu, Jan 26, 2023 at 12:31:06PM -0600, Justin Pryzby wrote:
>> On Thu, Jan 26, 2023 at 09:46:07AM -0500, Andrew Dunstan wrote:
>>> Would it be worth adding a check right at the end of the schedule that
>>> makes sure there are no such roles left?
>> Yes, because the alternative is to have cirrus run "installcheck"
>> twice..
> Agreed about doing something like that, with a new SQL script perhaps?
> We can rely on -DENFORCE_REGRESSION_TEST_NAME_RESTRICTIONS here, and
> I'd like to think that nobody creates such role names or they would
> see failures with the new query.
Yeah ... if we put it into an existing script, somebody will blindly
add more tests after it someday. I suggest calling it "test_cleanup"
to pair with "test_setup".
Is it worth checking for leftover regress_xxx tablespaces as well as
roles?
regards, tom lane