Re: possible bug in the 5.6.6 alter inplace software

Rich Prohaska <[email protected]> Tue, 10 Jul 2012 12:00:12 -0400
Newsgroups gmane.comp.db.mysql.devel
Message-ID <CAB3YYRVvMvDJov+qcZHzBccw3SU3Pw+gJZo9RiCS-dyO4d39KA@mail.gmail.com>
Hello Dmitry,
My example may have the same result (table copied) but the cause is
different (new columns are being added to the table, yet the fill
function sets the ADD and DROP flags).

On Tue, Jul 10, 2012 at 11:31 AM, Dmitry Lenev <[email protected]> wrote:
> Hello Rich!
>
> Thanks for finding and reporting this issue!
>
> * Rich Prohaska <[email protected]> [12/07/10 18:11]:
>> Hello,
>>
>> We found a alter table statement that adds columns to a table sets the
>> ADD and DROP INDEX flags in the alter inplace info struct.  When
>> running the attached mysql test program, innodb's
>> check_if_supported_inplace_alter gets called with unexpected
>> handler_flags set (handler_flags = 2115 = ADD_INDEX + DROP_INDEX +
>> ADD_COLUMN + ALTER_COLUMN_ORDER).  The fill_if_supported_inplace_alter
>> function is probably at fault.  We are using
>> mysql-5.6.6-labs-june-2012.
>
> Indeed, this is a bug in fill_alter_inplace_info() code.
> This function incorrectly determines prefix keys as dropped
> and then re-created by any ALTER TABLE. :(
>
> I have reported a bug about it: http://bugs.mysql.com/65865.
>
> Thanks once again!
>
> Best regards,
> Dmitry Lenev
>
> --
> Dmitry Lenev, Software Developer
> Oracle Development SPB/MySQL, www.mysql.com
>
> Are you MySQL certified?  http://www.mysql.com/certification

-- 
MySQL Internals Mailing List
For list archives: http://lists.mysql.com/internals
To unsubscribe:    http://lists.mysql.com/internals