Choosing a single layer of cleverness - how clever is it?

Hello Core :slight_smile:

If Davids intentions with single layer of cleverness is possible to
achieve (http://www.loudthinking.com/arc/000516.html), why did then
CDBaby.com need to change to PostgreSQL to be able to get control of
their data integrity, forced foreign keys and constraints? Was this
before Rails? (Ref.
http://www.oreillynet.com/ruby/blog/2005/11/migrating_to_ruby_on_rails_and.html)

Question for Jeremy Kemper: Is CDBaby using Rails with a single layer
of cleverness or do they still use the features of PostgreSQL RDBMS.
If CDBaby uses both functionallity from RDBMS and Rails model. How do
you handle this? Do you maintain both the database and Rails-models or
do you only configure one of the places?

Does Rails really work with a single layer of cleverness on
large/complex database applications or is it only intended on simpler
applications?

/MartOn