Re: Stupid Database Tricks

On 19 Jun 2004 16:11:47 -0700, --CELKO-- wrote:

>>> 0. we like to keep things familiar for the 60 year old COBOL
>programmers who wrote version 1. we keep each table in its own file.
><<
>
>I remember COBOL people doing this when DB2 came out! Actually, it is
>more like each former file is put into one SQL table. There are no
>relationships among the tables. I even saw the fields mapped into
>CHAR(n) or VARCHAR(n) instead of a temporal datatype or INTEGER, etc.
>
>Each READ became a FETCH so that the original COBOL logic did not have
>to change. Joins? We don't got no stinkin' Jions!

Hi Joe,

Less than two years ago, I was working for a major Dutch company. There was only one way allowed to access the DB2 tables: through an interface module that read one row at a time and returned the results to the calling application. Updates had to be done row-by-row as well, through another interface module. Joins were definitely not allowed!!! (I worked as a designer so I never got to see the actual program code, but I'm pretty sure it was all cursor-based)

I did try to convince them to change this habit, but it was all "company standard" - I lost the fight. It was already a major achievement that I did manage to convince people that interface module could also translate new structure to old structure, to reduce the changes in the rest of the app - a major achievement! Several people really hated my guts for daring to suggest that the parameters of an I/O module be actually any different than the fields (sic!) in the table.

Best, Hugo

-- 

(Remove _NO_ and _SPAM_ to get my e-mail address)