Date: Wed, 26 Sep 2018 11:11:27 -0300
From: Alvaro Herrera <alvherre@2ndquadrant.com>
To: Bruce Momjian <bruce@momjian.us>
Cc: "David G. Johnston" <david.g.johnston@gmail.com>,
	Michael Paquier <michael@paquier.xyz>,
	Andres Freund <andres@anarazel.de>,
	PostgreSQL-development <pgsql-hackers@postgresql.org>
Subject: Re: transction_timestamp() inside of procedures
Message-ID: <20180926141127.dvvkkms5gpuan3qc@alvherre.pgsql>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: <20180926004739.GA15378@momjian.us>
User-Agent: NeoMutt/20180716
Precedence: bulk

On 2018-Sep-25, Bruce Momjian wrote:

> Well, it is an entire paragraph, so it might be too much.  If you
> download the zip file here:
> 
> 	http://www.wiscorp.com/sql200n.zip
> 
> and open 5CD2-02-Foundation-2006-01.pdf, at the top of page 289 under
> General Rules, paragraph label 3 has the description.  It talks about
> procedure statements and trigger functions, which seems promising.

I have the 2011 draft, not the 2006 one; you seem to be referring to
<datetime value function> (which is 6.32 in the 2011 draft I have).
General rule 3 is entirely unreadable, and is followed by this note:

  WG3:LCY-025 took no action on the preceding instance of general containment.
  It was felt that this rule is too complicated, to the point of being virtually
  unintelligible. In addition, the rule does not recognize that <datetime value
  function>s can be evaluated implicitly as <default option>s. It is believed
  that this rule does not reflect actual practice and should be rewritten to
  align it with implementations. Note that Subclause 15.1, “Effect of opening a
  cursor”, also has a General Rule on this subject. See
  Possible Problem FND-992 .

In SQL2016, this rule was removed completely.

I don't think this offers any practical guidance.

-- 
Álvaro Herrera                https://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services