MySQL Bugs: #62453: Workbench deletes 4byte utf8 symbols without notification and inserts the string

Bug #62453	Workbench deletes 4byte utf8 symbols without notification and inserts the string
Submitted:	16 Sep 2011 20:33	Modified:	17 Sep 2011 9:49
Reporter:	denixx baykin	Email Updates:
Status:	Not a Bug	Impact on me:	None
Category:	MySQL Workbench	Severity:	S3 (Non-critical)
Version:	5.2.34	OS:	Windows (Win XP x32)
Assigned to:		CPU Architecture:	Any
Tags:	UTF-8, utf8, utf8mb4

Description:
Workbench deletes 4byte utf8 symbols without notification and inserts the string without 4byte symbols.
JDBC reports error when inserting.

How to repeat:
I'll attach file with instructions.

Here is an instructions. Run one at a time.

Attachment: Workbench utf8 4byte.txt (text/plain), 781 bytes.

I used mysql-5.5.15-win32.

you have utf8 as default charset for the database and create the table without engine specification. So the table/column will not handle 4 bit utf8 character.

But SQL_MODE has an influence here:

CREATE  TABLE `test`.`testutf8mb4` (
  `testValue` VARCHAR(50) NOT NULL ,
  PRIMARY KEY (`testValue`) );

USE `test`;

SET SQL_MODE = '';

INSERT INTO testutf8mb4 VALUES ('Сарнаут✔

2Peter Laursen
Please, add a text file with instructions, this mysql bug tracker does not allow to store 4byte symbols in messages, but in files it can.

Haha! This bugs system runs on MySQL5.1 and we cannot write 4 byte utf8 characters here! They are not getting saved and the complete post truncates exactly like in denixx's example!

The point is SQL_MODE:

SET SQL_MODE = ''; INSERT .. << truncates
SET SQL_MODE = 'strict_all-tables'; INSERT .. << returs error.

Peter
(not a MySQL person)

BTW: what say about running this bugs system in strict mode instead?

@Denixx: I just found out the same.  See my last post! :-)

Yes, with SET SQL_MODE = 'strict_all_tables'; it returns an error, thanks.

So, do you still think there is a bug here?

I think it's something like unexpected feature but still need a fixing because there is a difference in work of Workbench and a Connector/J.

Hm, now JDBC allows insertion of 4byte-symbols in table field but changes it to "?".
I slightly confused.
Maybe I  not tested it more than one time.

@Denixx: do you "SET NAMES utf8mb4"?  I think that both character_set_client and character_set_connection (at least) will have to be utf8mb4.

A private comment: originally in 5.5 early betas the old utf8 charset was renamed to utf8mb3 and utf8 was 4 byte. That was changed for some reason I don't know (maybe upgrade problems). But the first implemation of 4 byte utf8 was definitely the best.  What we have now is a mess.

No, all is okay.
Just dropped the schema and restarted the server.
Then rolled the changes like in example and JDBC shows an exception again.