Re: Overhauling GUCS - Mailing list pgsql-hackers

From David E. Wheeler
Subject Re: Overhauling GUCS
Date
Msg-id 866D369E-782F-4EBD-B4B4-250FF6068D0D@kineticode.com
Whole thread Raw
In response to Re: Overhauling GUCS  (Greg Smith <gsmith@gregsmith.com>)
List pgsql-hackers
On Jun 6, 2008, at 01:50, Andreas Pflug wrote:

> Two heretical questions:
> Do we need user generated comments at all?
> I can't remember ever having used any comment in postgresql.conf.

That's a valid point. I've used comments to note by whom and when when  
a setting was changed.

> Why do so many people here insist on editing postgresql.conf as  
> primary means of changing config params?
> Isn't a psql -c "SET foo=bar; MAKE PERSISTENT" just as good as  
> sed'ing postgresql.conf or doing it manually?

I think that there has been enough pushback against housing all the  
settings in the database, not to mention that it calls for an API,  
that just starting with something simpler to parse the file and  
rewrite it from the command-line might be a better first step.

> Looking around for different approaches, network appliances come to  
> my mind, e.g. Cisco routers and PIX. You have 3 ways to configure a  
> pix:
> - use a command line (using ssh or telnet, eqivalent to psql); WRITE  
> MEMORY to make the changes survive a reboot.
> - use a web interface (or similar tool)
> - use tftp to up/download the complete config in and out, editing  
> the file. User comments will be lost, with the exception of those  
> that have been applied with special comment commands (equivalent to  
> "comment on").

I think the first option there is the one that's been getting the most  
support here.

Best,

David



pgsql-hackers by date:

Previous
From: "David E. Wheeler"
Date:
Subject: Re: Overhauling GUCS
Next
From: Tom Lane
Date:
Subject: Re: Proposal: New LWLockmode LW_OWNER