Re: pg_ls_tmpdir to show directories and shared filesets (and pg_ls_*)

On Mon, Mar 14, 2022 at 01:53:54PM +0900, Michael Paquier wrote:
> +select * from pg_ls_logicalmapdir() limit 0;
> +select * from pg_ls_logicalsnapdir() limit 0;
> +select * from pg_ls_replslotdir('') limit 0;
> +select * from pg_ls_tmpdir() limit 0;
> +select * from pg_ls_waldir() limit 0;
> +select * from pg_stat_file('.') limit 0;
>
> The rest of the patch set should be stable AFAIK, there are various
> steps when running a checkpoint that makes sure that any of these
> exist, without caring about the value of wal_level.

I was contemplating at 0002 this morning, so see which parts would be
independently useful, and got reminded that we already check the
execution of all those functions in other regression tests, like
test_decoding for the replication slot ones and misc_functions.sql for
the more critical ones, so those extra queries would be just
interesting to check the shape of their SRFs, which is related to the
other patches of the set and limited based on my arguments from
yesterday.

There was one thing that stood out though: there was nothing for the
two options of pg_ls_dir(), called missing_ok and include_dot_dirs.
missing_ok is embedded in one query of pg_rewind, but this is a
counter-measure against concurrent file removals so we cannot rely on
pg_rewind to check that. And the second option was not run at all.

I have extracted both test cases after rewriting them a bit, and
applied that separately.
--
Michael