[thelist] Architecture for arbitrary data management

[Joe Flintham]

r937 wrote:
> sounds like you can encapsulate/instantiate/disambiguate (choose the 
> programming terminology du jour here) the template specs into a good old sql 
> CREATE TABLE statement, such that each template is a separate table, with 
> regular old columns, with specific column names and datatypes...
>
> shouldn't be all that hard to accomplish
>
>   
agreed, but thinking ahead to a year ahead, when there are, say 30 such 
tables, built around their corresponding templates:

say 8 of those 30 tables contain fields such as 'latitude' and 
'longitude', and the users wish to build an app in, say, Google Earth, 
using those 8 specific objects.  I can build a query based on the 
template definitions (probably stored in XML files), which will then 
return the data from 8 different tables.

Is that still better than going for the short term cost of storing all 
latitude and longitude data in their own table from the get-go?  (And by 
extension, all the other attributes they may wish to build on down the 
line...).  This latter route would effectively write the template 
definitions into the relational nature of the tables themselves.

Thanks again :)

Joe