Horton

Re: [PATCH] Stop ALTER SYSTEM from making bad assumptions

Tom Lane <tgl@sss.pgh.pa.us>

View thread View on postgr.es
From: Tom Lane <tgl@sss.pgh.pa.us>
To: Isaac Morland <isaac.morland@gmail.com>
Cc: Stephen Frost <sfrost@snowman.net>, Tomas Vondra <tomas.vondra@2ndquadrant.com>, Amit Kapila <amit.kapila16@gmail.com>, Ian Barwick <ian.barwick@2ndquadrant.com>, Robert Haas <robertmhaas@gmail.com>, "pgsql-hackers@postgresql.org" <pgsql-hackers@postgresql.org>
Date: 2019-08-05T15:05:07Z
Lists: pgsql-hackers 
Isaac Morland <isaac.morland@gmail.com> writes:
> Here's a radical suggestion: replace postgresql.auto.conf with a directory
> containing multiple files. Each file is named after a configuration
> parameter, and its content is the value of the parameter.

Hmm ... that seems like a lot of work and code churn --- in particular,
guaranteed breakage of code that works today --- to solve a problem
we haven't got.

The problem we do have is lack of documentation, which this wouldn't
in itself remedy.

> In order to prevent confusing and surprising behaviour, the system should
> complain vociferously if it finds a configuration parameter file that is
> not named after a defined parameter, rather than just ignoring it.

As has been pointed out repeatedly, the set of known parameters just
isn't that stable.  Different builds can recognize different sets of
GUCs, even without taking extensions into account.

			regards, tom lane

Commits

  1. Doc: improve documentation about postgresql.auto.conf.

  2. Fix ALTER SYSTEM to cope with duplicate entries in postgresql.auto.conf.