Aaron Trevena wrote:
> it's the scattering of SQL all over the place that requires changing
> by hand when the schema is updated or code is refactored, etc.

I'm curious, as a non-user of these classes - how does a schema change
not get reflected in how you have to call the classes?

It seems to me that simple schema changes like adding a column or
changed field sizes don't require any change to SQL either - if you
wrote it sensibly.  Similarly, classes which do some abstraction will
also have some schema changes which don't affect how you call the
classes, and some where you do have to update those calls all over
your code - same problem as Aaron says is avoided.

Seems to me that the main protection against this is to keep the
database calls in one module in your application - with that module's
API matching the logic of the application as cleanly as possible.

I'm not seeing how whether that app module uses some sQL, or various
abstraction classes, makes all that much difference to the work
involved when schema is changed.  But I'm sure it does.  Can someone
enlighten me?

Cheers :-)
-- Jamie
_______________________________________________
BristolBathPM mailing list
[email protected]
http://mailman.bristolbath.org/mailman/listinfo/bristolbathpm

Reply via email to