Re: Generalised approach to storing address details

From: Neo <neo55592_at_hotmail.com>
Date: 7 Dec 2006 08:35:05 -0800
Message-ID: <1165509305.776481.182560_at_79g2000cws.googlegroups.com>


> I want to set up a generic way of entering names and address details...
> It is quite possible that it is a well-known 'bad thing
> to do' (it feels a bit like EAV).

Some advantages of using EAV design in RMDBs (also pointed out by Cimode):
1) Flexibility to handle data, including that which is unknown at design time.

Some disadvantages of using EAV design in RMDBs: 1) Loss of SQL support resulting in more and complex code. In addition, more constraints have to be implemented via user code. 2) Decreased performance due to increased joins.

In general, pure EAV is possible but mostly impractical in RMDBs. You can begin to see the complexity of managing EAVs in recent thread titled "How to find Brothers and Sisters?"

Experiences with EAV in RMDBs led me to develop dbd, a small experimental database. It has the advantages of EAV while replacing SQL with a LISP-like interface. See following examples:

Persons with Varied Attributes:
www.dbfordummies.com/Example/Ex121.asp

Baseball Team Members with Varied Attributes: www.dbfordummies.com/Example/Ex102.asp Received on Thu Dec 07 2006 - 17:35:05 CET

Original text of this message