Re: EAV (Re: Object-relational impedence)

On Mar 25, 11:45 am, "David Cressey" <cresse..._at_verizon.net> wrote:
> "topmind" <topm..._at_technologist.com> wrote in message
>
> news:a39a2bfd-de5a-4aac-ae2a-e942e99006b8_at_d4g2000prg.googlegroups.com...
>
> > If the customer wants the ability to add new columns without
> > programmer or DBA intervention, we should have technology that can
> > deliver that. "It's not good for you" is NOT an acceptable answer.
>
> My response to this is from the mudpits of practice, and NOT from the ivory
> towers of theory.
>
> WHENEVER a customer demands the ability to add new columns without
> programmer or DBA intervention, the same customer will ALWAYS demand a
> reporting capability that can only be delivered by managing the data and
> keeping the data definitions under control.

With web-based reports, I find it's fairly easy to have dynamic columns if the customer realizes that wrapping etc. is generally by the decision of the browser, not developer. However, paper-centric reports are a different matter because they tend to be hand-tuned to cram as much on a page as possible because thats how customers tend to want it -- Single Page Syndrome.

>
> It doesn't matter how many times you explain to the customer that if he can
> add columns to the database at will, the semantics of those columns are
> beyond the knowledge of the existing reports. You can explain this early in
> the game or late in the game. It doesn't matter. The customer will ALWAYS
> hold you at fault for not being able to deliver the same capability with
> dynamically redefined data that you could formerly deliver with controlled
> data.
>
> So, if you can figure a way to hold your customers accountable for the
> consequences of their own actions, be my guest. But you have been warned:
> beyond here there be dragons.

I agree that some kind of CYA usually applies. The politics of software design can be challenging.

-T- Received on Wed Mar 26 2008 - 00:04:48 CET