On Thu, 29 Dec 2005 11:53, [EMAIL PROTECTED] wrote:
> Having sunk countless hours into this and have very little to show for it,
> personally I'm convinced there must be a better way.

The way I took / am taking with mini-gnumed is as follows:

the problems / situation:
1.) fully normalized tables proved to be too unwieldly / slow for some 
purposes - e.g. selecting patients from a list box after entering a name, and 
you want some demograpohic details (name, age, dob, street, town ...) display 
in the selector box

2.) denormalized tables can be a pain in the butt when updating data & trying 
to keep data consistent, but are very easy for reading information quick

3.) simulating denormalized tables with updateable views is a bit slow for 
reading purposes, and most access will be read-only

possible solution I took:

1.) start with denormalized tables to get things going quickly
Model the denormalized tables to the needs of the most commonly / frequently 
user interface elements to facilitate delay-less user interaction (= data 
model does not follow theoretical ideas about the nature of the data, but it 
follows the real world interaction the end user will have with the data)

2.) add normalized tables later, with trigger functions in both normalized and 
denormalized tables doing the housekeeping - wherever needed and appropriate

it works and is as fast as it gets, but it is not aesthetically pleasing and 
not scalable.

Horst
_______________________________________________
Gpcg_talk mailing list
[email protected]
http://ozdocit.org/cgi-bin/mailman/listinfo/gpcg_talk

Reply via email to