| Bug #13344 | CAST(1E+300 TO signed int) on little endian CPU, wrong result | ||
|---|---|---|---|
| Submitted: | 20 Sep 2005 9:55 | Modified: | 20 Oct 2005 18:20 |
| Reporter: | Matthias Leich | Email Updates: | |
| Status: | Closed | Impact on me: | |
| Category: | MySQL Server | Severity: | S2 (Serious) |
| Version: | 4.1 | OS: | |
| Assigned to: | Jim Winstead | CPU Architecture: | Any |
[20 Sep 2005 12:50]
Matthias Leich
Please forget the I assume we have this inconsistent behaviour since the views are available. It was a mistake during cut and paste.
[22 Sep 2005 22:55]
Bugs System
A patch for this bug has been committed. After review, it may be pushed to the relevant source trees for release in the next version. You can access the patch from: http://lists.mysql.com/internals/30240
[13 Oct 2005 15:13]
Jim Winstead
Fixed in 4.1.16 and 5.0.15.
[20 Oct 2005 18:20]
Paul DuBois
Noted in 4.1.16, 5.0.15 changelogs.
Description: Example: My box (Intel Pentium M - little endian) SELECT 1.0E+300, CAST(1.0E+300 AS SIGNED INT); 1.0E+300 CAST(1.0E+300 AS SIGNED INT) 1e+300 -9223372036854775808 <------ wrong BTW: INSERT INTO ... <SIGNED BIGINT> = 1.0E+300; Produces a warning + shows the expected value 9223372036854775807 That means the problem seems to be within the function CAST. aix52 (Power PC -- big endian) SELECT 1.0E+300, CAST(1.0E+300 AS SIGNED INT); 1.0E+300 CAST(1.0E+300 AS SIGNED INT) 1e+300 9223372036854775807 <----- correct The manual says in chapter 11.1.1. Overview of Numeric Types: BIGINT[(M)] [UNSIGNED] [ZEROFILL] A large integer. The signed range is -9223372036854775808 to 9223372036854775807 11.2. Numeric Types ... When asked to store a value in a numeric column that is outside the column type's allowable range, MySQL clips the value to the appropriate endpoint of the range and stores the resulting value instead. This behaviour on storage time is also applied to queries, CAST whenever the maximum value range of a numeric data type is exceeded. My environment: - Intel PC with Linux(SuSE 9.3) - MySQL compiled from source Version 4.1 ChangeSet@1.2417.1.2, 2005-09-11 Version 5.0 ChangeSet@1.1980, 2005-09-19 aix52: - PowerPC - Version 5.0 last ChangeSet around 2005-09-13 I assume we have this inconsistent behaviour since the views are available. This bug might be not so important for customers, but it harms the development of testcases. Currently we can only have one file with expected testresults for all hardware architectures. How to repeat: Please execute the statements above.