Message-ID: <3B3952DB.A665D28E@zhengmai.com.cn>
Date: Wed, 27 Jun 2001 11:28:28 +0800
From: He Weiping <laser@zhengmai.com.cn>
MIME-Version: 1.0
To: Tom Ivar Helbekkmo <tih@kpnQwest.no>
CC: Tom Lane <tgl@sss.pgh.pa.us>, pgsql-docs@postgresql.org
Subject: Re: doc question about column name alias
References: <3B374CF6.C1AB82F0@zhengmai.com.cn>
	<15864.993483354@sss.pgh.pa.us> <867kxzjmat.fsf@athene.i.eunet.no>
Content-Type: text/plain; charset=gb2312
Content-Transfer-Encoding: 7bit

Tom Ivar Helbekkmo wrote:

> Tom Lane <tgl@sss.pgh.pa.us> writes:
>
> >> SELECT a AS b FROM table1 ORDER BY a;
> >> I think it should be:
> >> SELECT a AS b FROM table1 ORDER BY b;
> >
> > You are correct that the latter is the SQL-standard version.
>
> In fact, this becomes pretty obvious when you consider the theoretical
> ordering of the evaluation of the various parts of the statement.  The
> ORDER BY is executed *after* the SELECT part, so it only knows about
> the column name 'b', and doesn't know what 'a' refers to.  Actually,
> ORDER BY isn't even part of the SELECT statement, really, but hides an
> implicit CURSOR operation taking place during data output.  Allowing
> the first version to work is of questionable value, since it might
> cause errors and confusion when code is modified.  Consider the case
> where you have "SELECT a AS b, b AS a ...".  I've done it!  :-)
>
> The order of evaluation is: FROM, WHERE, GROUP BY, HAVING, SELECT,
> and, finally, ORDER BY.

Thanks all for these info. & examples, they help me much to clearify
the question, the present wording is a little bit hard to understand, but
accurate and
efficent, and I think it's better to add a few lines to make it clear, like

"...but the SQL standard require (or "only permit") us to do:
select a as b from table1 order by b;"

    regards    laser