Re: pg_config, pg_service.conf, postgresql.conf ....

"Mark Woodward" <pgsql(at)mohawksoft(dot)com> writes:
> If one can specify a different port than the default on the command line,
> why wouldn't a file designed to describe the server process include it. My
> intention is to include all the options available via environment or
> command lon in the file.

I think that's a really bad goal. In the first place, we generally
discourage people from setting options on the postmaster command line
if they could be set in postgresql.conf, because if you set them on
the command line they can't be overridden later via editing the conf
file and SIGHUP. (Of course that doesn't matter for startup-time-only
options, but the fact remains that this isn't and shouldn't be the
preferred way to set options.) In the second place, the set of options
changes from version to version, so you'd be creating a large make-work
project for yourself trying to keep the config file parser in sync with
the available options. I think it's sufficient if you provide a single
config file entry which is extra command line options, eg,

SWITCHES="--port=5444 -i -F"

ISTM that postmaster pathname, data directory, and extra-switches are all
you really need. (Breaking out data directory is useful because pg_ctl
has its own reasons for wanting to know that, but it does not care about
the other switches.)

regards, tom lane