MySQL Bugs: #54212: 3.51 Driver returns wrong results about column sensitivity

Bug #54212	3.51 Driver returns wrong results about column sensitivity
Submitted:	3 Jun 2010 17:24	Modified:	18 Jun 2010 15:03
Reporter:	Lawrenty Novitsky	Email Updates:
Status:	Closed	Impact on me:	None
Category:	Connector / ODBC	Severity:	S3 (Non-critical)
Version:	3.51	OS:	Any
Assigned to:	Lawrenty Novitsky	CPU Architecture:	Any
Tags:	SQLColAttribute SQL_DESC_CASE_SENSITIVE

Description:
SQLColAttribute(...SQL_DESC_CASE_SENSITIVE...) returns SQL_FALSE for binary types and SQL_TRUE for the rest. Should be the other way around.

http://msdn.microsoft.com/en-us/library/ms713558%28v=VS.85%29.aspx
"SQL_TRUE if the column is treated as case-sensitive for collations and comparisons.
SQL_FALSE if the column is not treated as case-sensitive for collations and comparisons or is noncharacter."

Code
case SQL_DESC_CASE_SENSITIVE:
            *(SQLINTEGER *)NumericAttributePtr= (field->flags & BINARY_FLAG ?
                                                 SQL_FALSE : SQL_TRUE);

How to repeat:
the bug is pretty obvious from the code snippet

Suggested fix:
Make it in sync with 5.1 code

 if (field->flags & BINARY_FLAG) /* TODO this doesn't cut it anymore */
      irrec->case_sensitive= SQL_TRUE;
 else
      irrec->case_sensitive= SQL_FALSE;

Although the comment says that is not all what is needed there, it's definitely better than current code.

The described patch sounds fine. Consider it pre-approved by me.

The get_sql_data_type() function has what may be a better test for the binary-ness of a field.

the patch

Attachment: bug54212.diff (application/octet-stream, text), 3.47 KiB.

Jim, I have to ask you to re-review the patch since it's slightly different from what we discussed.
The difference is that c2(41th) field from the testcase wasn't recognized as binary by the "binary-ness test" from get_sql_data_type(). So i've slightly changed it.

pushed as rev#714. Probable destination is 3.51.28
I assume this approval as a blessing to apply similar change in other places of 3.51 and 5.1(about detecting binary field). but will probably do some more research on that.

An entry has been added to the 3.51.28 changelog:

SQLColAttribute(...SQL_DESC_CASE_SENSITIVE...) returned SQL_FALSE for binary types and SQL_TRUE for the rest. It should have returned SQL_TRUE for binary types, and SQL_FALSE for the rest.