Re: How is it possible to have invalid time in a time field?

This is where I get a little miffed with IBM. I thought that by using date and time data types we had some assurance of data integrity, so that situations as you describe would not occur. We can't add 30 to 20070901 and get 20070931. We have to add %days(30) and it gives us a valid answer. But apparently that isn't so with *hival.


Paul Nelson wrote:

Heh. I know a package that throws garbage into its files when calculating
due dates. If the invoice date is 9/1/07 and the terms are 30 days, one ends
up with 9/31/07 as the due date.

The vendor's response? Yeah, we know, but we code around it in our other
programs. Unfortunately, they store dates as YYYYMMDD.

Paul Nelson
Cell 708-670-6978
Office 512-392-2577
nelsonp@xxxxxxxxxxxxx


-----Original Message-----
From: midrange-l-bounces@xxxxxxxxxxxx
[mailto:midrange-l-bounces@xxxxxxxxxxxx] On Behalf Of rob@xxxxxxxxx
Sent: Friday, November 02, 2007 10:21 AM
To: Midrange Systems Technical Discussion
Subject: Re: How is it possible to have invalid time in a time field?

Tommy,

What you said holds some water, but when it is all said and done, you still need the Fort Knox of data validation. I've discovered that many packages do fine validation. It's the programmers out at the customer who are most likely to hose up the data. They are the ones who do data conversions, interface data from other batch processes, such as EDI or barcoding. Merging of databases, uploading from Excel, etc.

For example one package uses no primary key. And on their DDS for their logical they specify no UNIQUE key. If you use their program logic then the odds are slim to none that you will ever have duplicate keys. However in the process of merging databases and/or conversions we do. And that's in the item master. All this could have easily been stopped at the moat by a primary key constraint.

Rob Berendt