MySQL Bugs: #41442: Support ieee754 by supporting INF and NaN

Bug #41442	Support ieee754 by supporting INF and NaN
Submitted:	12 Dec 2008 18:57	Modified:	14 Dec 2008 14:27
Reporter:	Mark Callaghan	Email Updates:
Status:	Verified	Impact on me:	None
Category:	MySQL Server: Data Types	Severity:	S4 (Feature request)
Version:	5.0	OS:	Any
Assigned to:		CPU Architecture:	Any
Tags:	double, FLOAT, ieee754, inf, nan

Description:
MySQL does not support INF and NaN because the SQL spec doesn't allow that. If that is the case the spec is broken in this regard and makes SQL useless for scientific apps as INF (+INF and -INF) and NaN are expected values.

At the XLDB conference representatives from the large scale physics projects that generate too much data talked about the need to use databases to manage their metadata. If that metadata includes FP values, then they might not be able to use MySQL.

Oracle has great support for ieee754.

And MySQL still allows INF in some cases using 5.0
mysql> select 1.0e308 * 1.0e10;
+------------------+
| 1.0e308 * 1.0e10 |
+------------------+
|              inf |
+------------------+

But not with 5.1.30
mysql> select 1.0e308 * 1.0e10;
+------------------+
| 1.0e308 * 1.0e10 |
+------------------+
|             NULL |
+------------------+

When sql_mode='', +INF is rounded to MAX_DOUBLE in 5.0. It is NULL regardless of sql_mode in 5.1.

How to repeat:
see above

Bug #51217 was marked as a duplicate of this one.

Related:
Bug #57519 	Documentation for IEEE 754 compliance is missing