Bug #10897 FLOAT and DOUBLE obeying precision and scale. No documentation.
Submitted: 27 May 2005 1:33 Modified: 15 Aug 2005 19:07
Reporter: Dean Ellis
Status: Closed
Category:Server Severity:S3 (Non-critical)
Version:4.1.13 OS:
Assigned to: Alexey Botchkov Target Version:

[27 May 2005 1:33] Dean Ellis 
Description:
In 4.1, FLOAT and DECIMAL are obeying (precision,scale) settings, which is behaviorally
different than 3.23 and 4.0, where these were only display width and decimal precision. 
The behavior change, if intentional in 4.1, does not appear to be documented anywhere
(upgrade notes, either of the numeric type chapters) in the manual.

How to repeat:
DROP TABLE IF EXISTS t1;
CREATE TABLE t1 ( a FLOAT(5,2) );
INSERT INTO t1 VALUES (12345678.12345678);
SELECT a FROM t1;

4.0:

+-------------+
| a           |
+-------------+
| 12345678.00 |
+-------------+

4.1:

+--------+
| a      |
+--------+
| 999.99 |
+--------+

Suggested fix:
n/a
[6 Aug 2005 0:11] Mike Hillyer 
We need a three-part version number this applies to to properly document it with a
changelog entry and a manual update. Please provide three part version number. Setting
back to Open status.
[8 Aug 2005 14:57] Alexey Botchkov 
Pushed by VVA into 4.0.8
[15 Aug 2005 19:07] Mike Hillyer 
Documented in 4.0.8 changelog:

<listitem><para>The <literal>FLOAT</literal> and <literal>DECIMAL</literal> types now
obey (precision,scale) settings. (Bug #10897)</para></listitem>