Re: Some Laws

"Marshall Spight" <mspight_at_dnai.com> wrote in message news:sAC3d.76182$MQ5.9545_at_attbi_s52...
> Yeah, okay. Sometimes I make over-broad claims for shock value.

Yeah, so do I.

>
> But I still think that the category of applications that could benefit
from
> a DBMS is much larger than the category that actually uses them, and
> I think this is an artifact of the heavy early commercialization of
> the RM vs. other ways of doing things. (How many commercial
> "tree shaped file" manager packages can you name? How many
> non-commercial RDBMSs can you name?) (In other words,
> Larry Elison is to blame. :-)

Well, I'm also concerned about the number of applications that decided to use a DBMS, and then launched
their product on top of a truly crappy database design. One that was done in total ignorance of even first normal form, and without regard for any benefits of clarity in design whatsoever. Just completely spaghetti. Then, after they got terrible performance, in addition to low maintainablity, they decided that "the RDM is at fault".

I think a very large amount of harm has been done by this practice, both to software vendors themselves, and also to programmers, clients, and end users. I know this argues against my position that anyone should be allowed to attempt a database design. It's just that the recognition and rejection of crappy designs ought to be far more widespread.

> Photoshop could really use an RDBMS. Any online calendar
> program could really use an RDBMS. Anything that needs to
> make atomic updates of data structures could really use an
> RDBMS.
>

Well, they don't all need concurrent update control. And they don't all need redo logs.
And they don't all need to represent many:many relationships. Received on Mon Sep 20 2004 - 21:29:08 CEST