Re: alter enum add value if not exists

Bruce Momjian <bruce@momjian.us>

From: Bruce Momjian <bruce@momjian.us>
To: Tom Lane <tgl@sss.pgh.pa.us>
Cc: Hannu Krosing <hannu@2ndQuadrant.com>, Andrew Dunstan <andrew@dunslane.net>, Magnus Hagander <magnus@hagander.net>, PostgreSQL-development <pgsql-hackers@postgresql.org>
Date: 2012-09-26T19:12:24Z
Lists: pgsql-hackers
On Sat, Sep 22, 2012 at 06:08:19PM -0400, Tom Lane wrote:
> Hannu Krosing <hannu@2ndQuadrant.com> writes:
> > On 09/22/2012 11:49 PM, Andrew Dunstan wrote:
> >> Not really, I guess we should for the sake of consistency, although TBH
> >> I find it just useless noise and rather wish we hadn't started the 
> >> trend when we did the first DROP IF NOT EXISTS stuff.
> 
> > Time for a GUC
> > existence_notice = none | exists | not_exists | all
> 
> Not another one :-( ... isn't client_min_messages good enough?
> 
> We sort of had this discussion before w.r.t. the notices about creating
> primary key indexes etc.  I wonder whether we should make a formal
> effort to split NOTICE message level into, say, NOTICE and NOVICE
> levels, where the latter contains all the "training wheels" stuff that
> experienced users would really rather not see.  Or maybe just redefine
> NOTICE as meaning novice-oriented messages, and push anything that
> doesn't seem to fit that categorization into another existing message
> level?

I have always wanted a "novice" level, so we could warn about things
like unjoined tables.

-- 
  Bruce Momjian  <bruce@momjian.us>        http://momjian.us
  EnterpriseDB                             http://enterprisedb.com

  + It's impossible for everything to be true. +