Port MySQL patch for functions with constant values

From Jay:

Basically, the fixes are to ensure items in the parsed syntax tree which evaluate to a constant (such as UNIX_TIMESTAMP(constant) are only evaluated once and then cached in the Lex as a Item_cached_xxx object.

The port would be to take the commit and apply it almost as-is to Drizzle, add the associated test cases, and cleanup any non-Drizzle stuff in the commit (pretty minimal).

First come, first served! I can help out any takers on IRC.

-------- Original Message --------
Subject: Re: Functions of constant values
Date: Thu, 03 Dec 2009 18:53:54 +0100
From: Sergei Golubchik <email address hidden>
To: Olivier <email address hidden>
CC: <email address hidden> <email address hidden>
References: <email address hidden> <email address hidden> <email address hidden> <email address hidden> <email address hidden> <00af01ca7371$3687d250$a39776f0$@com> <email address hidden> <email address hidden>

Hi, Olivier!

> Is there a reason to compute some functions of constant values multiple times?

No.

> While debugging, I found unix_timestamp(constant) gets evaluated 3
> times in the (silly) example below (MySQL 5.1.41):
>
> DROP DATABASE IF EXISTS `test`;
> CREATE DATABASE `test`;
> DROP TABLE IF EXISTS `test`.`test`;
> CREATE TABLE `test`.`test` (
> `t` timestamp NOT NULL DEFAULT CURRENT_TIMESTAMP ON UPDATE CURRENT_TIMESTAMP
> ) ENGINE=InnoDB DEFAULT CHARSET=utf8;
> INSERT INTO `test`.`test` VALUES('2009-12-03 14:00:00'),('2009-12-03 14:01:00'),('2009-12-03 14:02:00');
> SELECT * FROM `test`.`test` WHERE unix_timestamp(t) >= unix_timestamp('2009-12-03 14:01:00');
>
> I thought deterministic functions of constant values where always
> optimized away by the query engine? Or this is a bug in
> unix_timestamp()?

The fix was just pushed this week.
http://bugs.mysql.com/33546
http://bugs.mysql.com/47839
http://bugs.mysql.com/41153

It didn't make it into 5.1, but it'll be in the next GA release.

--