Re: Proposal for Allow postgresql.conf values to be changed via SQL - Mailing list pgsql-hackers

On Tuesday, November 06, 2012 11:30 PM Robert Haas wrote:
> On Wed, Oct 31, 2012 at 8:17 AM, Magnus Hagander <magnus@hagander.net>
> wrote:
> >> I'm not convinced we ever *had* a consensus on this.  There were
> >> proposals, but I'm not sure a majority ever bought into any one of
> 'em.
> >
> > I thought there was a consensus. But given that the one I thought we
> > had consensus on was different, I'm not sure we can correctly call it
> > consensus.
> >
> > What we discussed at that time was to have a *function* that changes
> > the permanent configuration, and not actually extend the syntax of the
> > system. As a starting point.
> >
> > The idea at the time was to use the include *directory* functionality,
> > for say a "config.d" directory in pgdata. The builtin one would then
> > use a predictable filename in this directory, so that the DBA who
> > prefers it can drop files both before and after that file into the
> > directory.
> 
> Reading over this thread, it seems that there are at least three
> different proposals for how this should work in detail:
> 
> 1. Have a configuration file that can be rewritten using SQL, and have
> postgresql.conf include it by default.
> 2. Have a configuration directory that gets included in
> postgresql.conf by default, and one file in that directory will
> contain all the parameters set via SQL.
> 3. Have a configuration directory that gets included in
> postgresql.conf by default, with one file per parameter, and rewrite
> just that file when the corresponding parameter is set via SQL.
> 
> Also, there are at least three different proposals for what the syntax
> should look like:
> 
> 1. ALTER SYSTEM
> 2. SET PERSISENT
> 3. pg_frob_my_configuration()

This is very good summarization of all discussion in this mail chain.
However there is one more point which I am not able to clearly make out is
how to write into file that contains
all configuration parameters changed by SQL.
What I could understand from Greg and Josh's mail is that they are
suggesting to write a file by collecting active changed parameters from
memory or use pg_settings.
But as mentioned in other mail as per my understanding that this can lead to
have incorrect values in .auto file.
I think I am missing or not able to understand how can it be done without
reading .auto file or by communicating with other backends?
Can you please point me what is wrong in my understanding?

> For all of that, I think there is broad agreement that being able to
> set configuration parameters on a server-wide basis via SQL commands
> is a useful thing to do.  I certainly think it is.

> It seems to me that the only reason why we have any of this
> information in a text file at all is because there are some parameters
> that the server has to know before it can start.  After all, ALTER
> USER and ALTER DATABASE and ALTER FUNCTION store their values in the
> database itself, and no one has said, oh, those values should be
> stored in a file so people can edit them with a text editor.  Why not?
>  Surely that's just as defensible as wanting to edit the server-wide
> parameters, but you can't.