MySQL Bugs: #19345: Table column length multiplies on size session character set

Bug #19345	Table column length multiplies on size session character set
Submitted:	25 Apr 2006 17:25	Modified:	3 Aug 2007 10:13
Reporter:	Vladimir Loubenski	Email Updates:
Status:	Closed	Impact on me:	None
Category:	Connector / ODBC	Severity:	S2 (Serious)
Version:	3.51.12	OS:	Windows (Windows XP)
Assigned to:	Jim Winstead	CPU Architecture:	Any

Description:
Session characterset has impact on result execution ODBC function SQLColumn.
If session character set is default it works correct. If session character set will be changed to utf8 (On tab "Connect Option" -  "Initial statement" write: SET NAMES 'utf8'). It multiplies by 3 the value for COLUMN_SIZE.

How to repeat:
1. CREATE TABLE T_LENGTH (F11 VARCHAR(10))
2. Create DSN connection with default session charset
3. Call function SQLColumns to verify column length. It returns correct length 10 for COLUMN_SIZE
4. Change DSN, set session charset to utf8 (On tab "Connect Option" -  "Initial statement" write: SET NAMES 'utf8')
5. Repeat p3.  It returns length 30.

According to ODBC specification (http://msdn.microsoft.com/library/default.asp?url=/library/en-us/odbc/htm/odbcsqlcolumns.a...) reporter is right! COLUMN_SIZE is in characters, and it should be 10. Unlike BUFFER_LENGTH...

Hi Valeriy,
I’d like to clarify the problem.
I agree with you that COLUMN_SIZE should be in characters and result 10 is correct.
Problem is if in DSN  for ODBC connection I set session characterset to utf8 (On tab "Connect Option" -  "Initial statement" write: SET NAMES 'utf8') I get COLUMN_SIZE 30 for the same table.

Regards,
Vladimir.

Please use SQLColumns functions from unicode library (A,W). It is obvious why you have 3 times bigger result with non-unicode function just by counting bytes needed for unicode...

Thank you Tonci for answer.
I got the same result with using W function.
Put attention on answer for   #12894

Vladimir, can you post your test case, presumeably in C++?

Table t_length should be created

Attachment: mainWMySQL.cpp (text/plain), 3.36 KiB.

Hi, Vladimir. I don't even get connected with your code. I'll write it myself and post the findings.

Verified as described by reporter. Test case attached.

VS test file

Attachment: Testbug19345.cpp (text/plain), 4.01 KiB.

Hi Tonci,
I run your test case and got the same problem:
I get column length 30 on DSN which defines utf8 session character set.

You need adjust connection sections in my test case to be able connect to database :
 SQLWCHAR * pServer = (SQLWCHAR*)L"MySQLlocal";  ->
SQLWCHAR * pServer = (SQLWCHAR*)L" Your DSN";

SQLWCHAR * pUser = (SQLWCHAR*)L"rnddb6";- >
SQLWCHAR * pUser = (SQLWCHAR*)L"Your user name";

SQLWCHAR * pPassword = (SQLWCHAR*)L"rnddb6"; ->
SQLWCHAR * pPassword = (SQLWCHAR*)L"Your Password";

SQLWCHAR * pCatalog = (SQLWCHAR*)L"rnddb6"; ->
SQLWCHAR * pCatalog = (SQLWCHAR*)L"Your Database Schema name";

SQLWCHAR * pSchema = (SQLWCHAR*)L"";  //No changes

SQLWCHAR * pTable = (SQLWCHAR*)L"t_length";  //No changes

Thank you,
Vladimir.

Preliminary patch to clean up length handling, and other related bugs

Attachment: bug19345.patch (text/plain), 50.30 KiB.

Complete patch to fix column lengths and related bugs

Attachment: bug19345.patch (text/plain), 61.08 KiB.

The fix for this bug, and many related ones, has been committed, and will be in the 3.51.18 release.

Thanks for the bug report.

A note has been added to the 3.51.18 changelog: 

Obtaining the length of a column when using a character set for the connection of utf8 would result in the length being returned incorrectly.