Bullet Proof Design Disscussion

In general I am looking at discussing methods for handling error situations and recovering so I can complete a string of tests ending with a list of tests that passed and those that failed instead of stopping the test run on any error.  I know how to enable this feature and how to control it on a per KWT and per command basis.  The problem is error and event handling.  I do a little scripting but not much.  I have not implemented any sort of event or error handling as of yet.  I am working with an application with a database backend.  In many cases if a test fails it may leave the data in an odd state which will cause subsequent tests to fail, produce false positives or crash the app. 

The foundation will be a good name map.  My thoughts are to get all of my objects mapped and leave the name map as close to the object map as possible.  Then I would simplify the alias map as much as possible to make my keyword tests easier to read.  Only after this point would I being writing tests.

I plan on implementing some new approaches on my next project.  Here's to a productive discussion! 

Thanks all!