Re: pgsql: Support "postgres -C" with runtime-computed GUCs

On Thu, Sep 16, 2021 at 05:56:49PM -0400, Tom Lane wrote:
> Test case for this seems to have newline-related issues on Windows.

Thanks. I missed it.

> https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=fairywren&dt=2021-09-16%2018%3A36%3A14
>
> # Failed test 'data_checksums=on is reported on an offline cluster stdout /(?^:^on$)/'
> # at t/002_actions.pl line 184.
> # 'on
> # '
> # doesn't match '(?^:^on$)'

The regex should match AFAIK, and that worked in my own WIN32,
non-Msys, environment.

I see the business in PostgresNode::command_like_safe to work around
such a case like that in pg_ctl (efd7f8e), very similar to what I am
doing here so an extra chomp() should address this issue. Shouldn't
we try in the long-term to make the command_checks_* routines use
temporary files rather than the raw outputs if these are not reliable,
by the way? Adding Andrew in CC about that.

I am not completely sure what's going on here yet, so I'll just switch
the test to be skipped when Msys is involved for now. That should be
enough to bring back those machines to green.
--
Michael