List:General Discussion« Previous MessageNext Message »
From:Sasha Pachev Date:April 18 1999 5:24am
Subject:Re: auto_increment problem
View as plain text  
Michael Widenius wrote:
> 
> >>>>> "Lew" == Lew Barnesson <lewb@stripped> writes:
> 
> Lew> Hi,
> Lew> I began using auto_increment in an extensive system (46 'C' CGI programs on
> Lew> Win NT 4.0 server ) back in the 3.21.x days; back when I didn't know about
> Lew> the problem of a delete freeing up a previously used auto_increment-assigned
> Lew> number/id. My system assumes uniqueness accross not only the database, but
> Lew> archived deleted rows.
> 
> Lew> I understand this problem will be resolved in 3.23.0. Is this correct, and
> Lew> when can I expect to see this release? My system is going production very
> Lew> shortly, so I need to decide to wait for resolution, or to create my own
> Lew> unique id assignments.
> 
> Lew> My complements to Monty and crew for this very fine mysql.
> 
> Lew> Regards to all.
> 
> Lew> -Lew Barnesson
> 
> Yes, MySQL 3.23 will solve this problem.  I will also very shortly release an
> alpha version of this.  The problem is of course when the 3.23 version
> will be ready for production use. (We will of course test 3.23 very
> throughly, but its very hard to test the new ISAM as good as the old
> ISAM has been tested the past few years)
> 
> You can also in the MySQL manual find an example how to solve this
> problem by using the LAST_INSERT_ID() function.
> 
> Regards,
> Monty
> 

What Monty suggests is a cleaner coding practice. All that a defensive
programmer would expect from auto_increment is that the id is unique for
the table.

-- 
Sasha Pachev
http://www.sashanet.com
Thread
auto_increment problemLew Barnesson17 Apr
  • auto_increment problemMichael Widenius17 Apr
  • Re: auto_increment problemSasha Pachev18 Apr
    • Re: auto_increment problemMichael Widenius18 Apr