Re: pgsql: Refactor all TAP test suites doing connection checks

On Mon, Apr 05, 2021 at 08:15:31PM -0400, Tom Lane wrote:
> Dunno.  Certainly most animals that enable TAP at all should be
> running the "authentication" test, but those others all need to
> be manually enabled, I believe.  My own animals run the SSL test
> but not LDAP or kerberos.

The SSL tests have always sent down a query even for scenarios where
the connection would fail, but they have used command_fails_like()
rather than PostgresNode::psql.  The puzzle does not seem complete.

> I'm fairly sure that we've previously fixed this exact same issue
> in some tests that expected a connection failure ...

Yeah.  Looking at the git history, that would be c757a3da.
test_access() in the kerberos tests sends a SELECT true even for
expected failures, so it seems to me that it would fail with the same
symptoms on those machines.  That looks worth fixing and
backpatching.

And I need more caffeine this morning..  I missed your point that a
connection failure should pass down an undefined query string per
that.  Well, the fix is simple then now that there is a single code
path for all those connection attempts.
--
Michael