Re: Row locks in DB2/400 UDB

["Jim Franz" <franz400@xxxxxxxxxxxx>]

>in the holy name of the audit...
if you have ever been thru a fraud investigation, you would be glad
for the audit trail. however, i have yet to see a valid requirement for
assigning the order# before confirm/complete. if creating temp records
before complete, you need a key, but not the final order#. If your orders
need a later "approval" , then the order really does exist, and should not
be trashed if later "not approved".
jim

----- Original Message -----
From: "Jim Damato" <jdamato@xxxxxxxxxxxxxxxxx>
To: "'Midrange Systems Technical Discussion'" <midrange-l@xxxxxxxxxxxx>
Sent: Friday, August 08, 2003 11:33 AM
Subject: RE: Row locks in DB2/400 UDB


> I've seen apps where, in the holy name of the audit, programs have been
> changed to write canceled transactions as canceled transactions.  In order
> to preserve an unbroken sequence a record is written every time a next
> sequence number is generated.
>
> The database equivalent of "this page was intentionally left blank"
>
> -Jim
>
> -----Original Message-----
> From: rob@xxxxxxxxx [mailto:rob@xxxxxxxxx]
> Sent: Friday, August 08, 2003 9:56 AM
> To: Midrange Systems Technical Discussion
> Subject: Re: Row locks in DB2/400 UDB
>
>
> Booth,
>
> Why not skip numbers?
> Perhaps it has to do with the fact that some ISO auditors are real
> a$$holes.  Like the ones that we insist that we get the source code to MS
> products so that we can change it to say "Page x of y".  Lets them know if
> there are any missing pages.
>
> Having a server program is a good idea.  Powered by a data queue or some
> such thing.  If the backout concerns you, then do it as a trigger.  Only
> when it actually goes to write the record to disk do you grab a new
> number.
>
> Of course, if you're running V5R2 then I'd use an identity column.
>
>
> Rob Berendt
> --
> "They that can give up essential liberty to obtain a little temporary
> safety deserve neither liberty nor safety."
> Benjamin Franklin
>
>
>
>
>
> "Booth Martin" <Booth@xxxxxxxxxxxx>
> Sent by: midrange-l-bounces@xxxxxxxxxxxx
> 08/08/2003 09:11 AM
> Please respond to Midrange Systems Technical Discussion
>
>         To:     <midrange-l@xxxxxxxxxxxx>
>         cc:
>         Fax to:
>         Subject:        Re: Row locks in DB2/400 UDB
>
>
> How much time is there between the access and the update? the three lines
> of
> code should be together. "get the number", "add 1 to the number", "update
> the record".
>
> Why not write one program that does that, and let the other programs call
> that program for the next number?
>
> Sometimes people like to get the next number but not do the actual update
> till the user has finished the process just in case the user bails out.
> The
> idea being so that there's no wasted numbers. That always struck me as an
> odd decision because numbers are free.
>
>
>
> ---------------------------------------------------------
> Booth Martin http://www.MartinVT.com
> Booth@xxxxxxxxxxxx
> ---------------------------------------------------------
>
> -------Original Message-------
>
> From: Midrange Systems Technical Discussion
> Date: Friday, August 08, 2003 9:02:07 AM
> To: midrange-l@xxxxxxxxxxxx
> Subject: Row locks in DB2/400 UDB
>
> All
>
> Is there a way to lock a row in DB2/400 UDB SQL without using commitment
> control?
>
> We have a table that holds the 'next sequence number'. Whenever several
> processes access the table, we end up with a scenario like:
>
> program A accesses the table and grabs
> the next counter number (counter number = 50)
>
> program B accesses the table and grabs
> the next counter number (counter number = 50)
>
> program B adds one to the counter
> number, and updates the table (counter number = 51)
>
> program A adds one to the counter
> number, and updates the table (counter number = 51)
>
> Now, there are two transactions with an identical sequence number.
>
> We're doing a simple select to fetch the sequence number, and we are not
> using commitment control.
>
> Any suggestions?
>
> Thanks
>
> -Doc
>
> _______________________________________________
> This is the Midrange Systems Technical Discussion (MIDRANGE-L) mailing
list
> To post a message email: MIDRANGE-L@xxxxxxxxxxxx
> To subscribe, unsubscribe, or change list options,
> visit: http://lists.midrange.com/mailman/listinfo/midrange-l
> or email: MIDRANGE-L-request@xxxxxxxxxxxx
> Before posting, please take a moment to review the archives
> at http://archive.midrange.com/midrange-l.
>
>