Re: Getting rid of regression test input/ and output/ files

On Sun, Dec 19, 2021 at 5:48 PM Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:

> Corey Huinker <corey(dot)huinker(at)gmail(dot)com> writes:
> > I have a nitpick about the \getenv FOO FOO lines.
> > It's a new function to everyone, and to anyone who hasn't seen the
> > documentation it won't be immediately obvious which one is the ENV var
> and
> > which one is the local var. Lowercasing the local var would be a way to
> > reinforce which is which to the reader. It would also be consistent with
> > var naming in the rest of the script.
>
> Reasonable idea. Another thing I was wondering about was whether
> to attach PG_ prefixes to the environment variable names, since
> those are in a more-or-less global namespace. If we do that,
> then a different method for distinguishing the psql variables
> is to not prefix them.

+1 to that as well.

Which brings up a tangential question, is there value in having something
that brings in one or more env vars as psql vars directly. I'm thinking
something like:

\importenv pattern [prefix]

(alternate names: \getenv_multi \getenv_pattern, \getenvs, etc)

which could be used like

\importenv PG* env_

which would import PGFOO and PGBAR as env_PGFOO and env_PGBAR, awkward
names but leaving no doubt about where a previously unreferenced variable
came from.

I don't *think* we need it for this specific case, but since the subject of
env vars has come up I thought I'd throw it out there.