RE: Primary and Foreign Keys and Blueberries

["Bullard, Claude L (Len)" <clbullar@ingr.com>]

Title: RE: Primary and Foreign Keys and Blueberries

Whoa, NELlie!  The W3C can break that horse.

 

I'm just exploring the wrinkles of reverse engineering an XML Schema from

a relational db model where one wants the result to be implementation

independent (insofar as that is doable).  So one has to find all of the

shortcuts taken by the original data designer and find out if they

make the cut.  Otherwise, as soon as one goes to share the schema,

one is facing the room full of other designers whose data also is full of

shortcuts and both sides are claiming necessity as the Mother of

invention.   I also need to meet the head-on claims of the grey beards that

"XML Schema doesn't do anything, so don't bother; XML Schema is

just another presentation; don't bother" and be able to show them

what it does and how that is a better deal than CSV negotiations.  

Len
http://www.mp3.com/LenBullard

Ekam sat.h, Vipraah bahudhaa vadanti.
Daamyata. Datta. Dayadhvam.h

-----Original Message-----
From: Anderson, John [mailto:John@Barbadosoft.com]

"Bullard, Claude L (Len)" wrote:

> Another wrinkle:  going into the relational db and finding out
> that some fields and tables have the same name (noted as duplicate
> declarations in the XML Schema validation) and finding one that is
> different only the by the spelling of a single character but is
> now so deep in the code, getting rid of it would be expensive.
> For now, I will annotate and rename.  I realize I could push
> all the fields into attributes but I really don't want to.

However, (feel free to flame me if I'm wrong) I don't think there are any RDBMSs out there which support the full range of UNICODE characters, or indeed NELs, in table or column names, nor are they likely to in the near future, so at least whatever you do won't be broken by Blueberry.

John