Unless you neeed to perform queries on the different (partner-specific) parts of the submissions, there is no need to use different columns. I would imagine a single table containing fields submission_ID, partner_ID and submission_data, plus possibly other fields with a global (not partner-related) meaning. The data would be a blob in a structured format. You could optimize for parsing speed by choosing your own format or go with XML.

As to the modules, I would associate to each partner one or more modules, perhaps classes inheriting from some common superclasses. When retrieving and managing the submission, the partner-specific module would take care of converting the blob to an in-memory representation. Something analogous would happen when creating, serializing and storing the submission. The modules would be retrieved on-the-fly (with some caching mechanism in place) by either perl's standard file-based "require" system or by reading them from a DB.

As to the choice of putting perl code in a DB, it is not wrong and in many cases it's a very advisable choice. Consider that the everything engine powering PM does it. It lets you modularize your application and make it extensible in a clean way and makes online updates doable without the risks involved in messing with the application filesystem.

Antonio

The stupider the astronaut, the easier it is to win the trip to Vega - A. Tucket