Subject: Re: [COMMITTERS] pgsql: Add index entries for more functions
From: Peter Eisentraut <peter_e@gmx.net>
To: Thom Brown <thom@linux.com>
Cc: pgsql-docs@postgresql.org
In-Reply-To: <AANLkTinVncuzW-wFnDdF0J8eqFrP2aUwcTZwyiuTGH=-@mail.gmail.com>
References: <E1PL0xu-00027P-5M@gemulon.postgresql.org>
	<AANLkTinVncuzW-wFnDdF0J8eqFrP2aUwcTZwyiuTGH=-@mail.gmail.com>
Content-Type: text/plain; charset="UTF-8"
Date: Tue, 30 Nov 2010 19:55:48 +0200
Message-ID: <1291139748.13957.6.camel@vanquo.pezone.net>
Mime-Version: 1.0
Content-Transfer-Encoding: 8bit

On sön, 2010-11-28 at 23:59 +0000, Thom Brown wrote:
> Whilst looking at the latest build of the docs at
> http://developer.postgresql.org/pgdocs/postgres/bookindex.html I've
> noticed a discrepancy in this commit from the patch I provided.  I
> specified "average" as a primary term with "avg" as a secondary term,
> but the commit shows them both as separate primary entries.  Was this
> changed intentionally?

Yes.

I have no written reference for this, but from experience, my
understanding of the way index entries should work is that the primary
term is something you would look up, and the secondary term is something
you would look up among the primary terms.

So I think a reader would probably look up both "avg" because that's the
function he's using, and "average" because that's the term for what he's
trying to do.

Somewhat made up examples of what reasonable secondary terms might be:

avg,
   for user-defined types
   performance of