Re: How to create a DBMS from scratch?

From: hRodstein <dragoninbabylon_at_cox.net>
Date: Thu, 05 Dec 2002 21:37:14 GMT
Message-ID: <eGPH9.30670$K64.1166056_at_news2.west.cox.net>


How about an n-level commission structure calculation? (where n is in excess of 20 recursions) Something that falls apart in SQL very quickly. Recursive data relationship problems are "pert-near impossible. Remember, "relational" data bases are clumsy ways to associate flat data record images. SQL was developed so data collected by mainframes using COBOL could more easily be accessed with a 4 gl language. The basic model is STILL Holorith cards (developed in the late 19th century). The Pick/MV model came out of a TRW think-shop in the late 60's. Even Codd in his later papers praised the MV model and agreed that hist original "rules" for relational data bases was sorely inadaquate for real-world business applications. But, so much "main stream" money was pumped into the data bases based on flat tables and Codd's original rukes, he was promptly ignored.
In my experience, Pick shops that have moved from the mv model to "relational" data bases have less than 60% of the functionality of their mv System- 5-6 years after the fact However, the IT director can command a 300K salary. Great for IT, sucks for the business.

"Eric Junkermann" <eric_at_deptj.demon.co.uk> wrote in message news:70568f73.0212050538.78da25fc_at_posting.google.com...
> emmettwa_at_yahoo.com (Emmett) wrote in message
news:<92d3639d.0212041707.f1b767_at_posting.google.com>...
> >
> > It was a total disaster precisely because so much of what was
> > done in Pick couldn't be done at all using a relational DB.
> >
>
> ... by the people who were doing it! They had to design and test the
> new application and design and test the data conversion - someone got
> some or all of it wrong. There are enough failed software
> developments in the world that you can probably find one to apparently
> back up any x vs y argument.
>
> I'm afraid that if you want this one to carry any weight you'll have
> to dig up a genuine statement of something that was done in Pick that
> can't be done in a relational DBMS (or maybe just couldn't be done in
> Oracle at the time, since that is _not_ the same thing).
>
> Eric Junkermann
> -- beware of hammer syndrome!
Received on Thu Dec 05 2002 - 22:37:14 CET

Original text of this message