Scrolling through the list of currently available (May CTP) reference mode options, I'm wondering if a option for generic formatted strings would be useful. Creating a model to compare approaches in another thread, I sub-typed email addresses, phone numbers and postal addresses. I gave the super-type the RM of ".code" as that seemed like the best available fit. I figure the generated database would use a VarChar type.
What are your thoughts on sub-typing entities with RMs that would (as separate objects), have specific formatting - such as VIN "numbers" or phone numbers (both of which are actually character strings)? Is it better to have the single (if more generic), RM for the super-type, or separate more semantically expressive RMs for the subtypes? I tend to go with the more expressive conceptual approach, figuring that type validation can be done elsewhere. Of course, the given UofD and requirements have to be considered.