Bug #21010 GROUP BY clause resolves a string as an identifier

• Submitted: 12 Jul 2006 17:38
• Modified: 12 Jul 2006 20:34
• Reporter: Dean Ellis
• Status: Duplicate
• Category: MySQL Server
• Severity: S3 (Non-critical)
• Version: 4.0 4.1 5.0 5.1
• CPU Architecture: Any

[12 Jul 2006 17:38] Dean Ellis

Description:
GROUP BY clause mis-resolves a string as an identifier, when that string has the same value as the name of a column or alias in the query.

How to repeat:
CREATE TABLE t1 SELECT 1 a UNION SELECT 2;
SELECT * FROM t1 GROUP BY a;
SELECT * FROM t1 GROUP BY 'a';
SELECT * FROM t1 GROUP BY 'b';

Currently, the first two SELECTs produce identical result sets (two rows each).

Because 'a' is a string and not an identifier, the final two SELECTs should produce identical result sets (one row each).

Suggested fix:
Treat string constants as strings, whether or not they have the same value as an identfier name.

[12 Jul 2006 19:49] Martin Friebe

seems identical to #14019 ?

[12 Jul 2006 20:34] Dean Ellis

Duplicate of bug #14019.  Thanks Martin.