Do you use APEX built-in DML functionality or code it manually?

Question

Just out of interest more than anything, wondering what everyone else does, and what advantages/disadvantages they see in their method?
Personally I never use the built-in stuff for manipulating, I create all the items and then a PL/SQL procedure in a package. Then I code a process to call that procedure to do any inserts/updates/deletions. I know this is a long way around to do something which APEX does fairly well (for simple situations at least anyway) but I find it far more flexible to go with this route and I prefer anything which is changing data to live on the database (as an object).
It also means that (other than changing the columns etc) changes can be made a lot more easily - APEX version control is pretty poor for a production environment and team-coding so the fewer changes to the app the better.
Anyone else any views?

Answer 1

I pretty much always use packaged up stored procedures for DML operations. All business logic goes into the packages where possible - in fact, the only element where I break this encapsulation methodology is when defining reports - even then, I've sometimes implemented reports as a table function, although this approach can cause more problems than it solves, performance-wise. Views provide an acceptable degree of abstraction in those cases when I want to hide the complexity of a query.
It's pretty rare that I'll be interrogating and updating a single table, although I take Roel's point re: updatable views and "instead of" triggers.