Re: pg_createsubscriber TAP test wrapping makes command options hard to read.

On 2025-01-24 Fr 1:59 PM, Dagfinn Ilmari Mannsåker wrote:
> Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> writes:
>
>> Michael Paquier <michael(at)paquier(dot)xyz> writes:
>>> On Thu, Jan 23, 2025 at 08:25:45PM +0000, Dagfinn Ilmari Mannsåker wrote:
>>>> Here's a patch for that.
>>> Thanks. I had a bit of time today and applied it.
>> BF member drongo doesn't like the new test for amcheck.
>> Looks like it has to do with SSPI authentication producing
>> a different error than expected:
> Ah yes, I always forget about that quirk on Windows.
>
>> stderr:
>> # Failed test 'checking with a non-existent user stderr /(?^:role "no_such_user" does not exist)/'
>> # at C:/prog/bf/root/HEAD/pgsql/src/bin/pg_amcheck/t/002_nonesuch.pl line 86.
>> # 'pg_amcheck: error: connection to server at "127.0.0.1", port 19928
>> failed: FATAL: SSPI authentication failed for user "no_such_user"
>> # '
>> # doesn't match '(?^:role "no_such_user" does not exist)'
>> # Looks like you failed 1 test of 107.
>>
>> You might be able to work around this with auth_extra,
>> a la 1e3f461e8 and other past fixes.
> Here's a (blind, but at least doesn't break on Linux) patch for that.
>
> - ilmari

This seems to me like the wrong fix. We don't want to create
"no_such_user" I think, we just want to catch the Windows error message,
as in this patch.

cheers

andrew

--
Andrew Dunstan
EDB: https://www.enterprisedb.com