Is anyone using Test Driven Development to build python geoprocessing tools?

Discussion created by g3martin on Mar 2, 2012
As Arcpy gets more robust allowing for more complex scripts and tools it seems like my development processes should grow with it. I currently use a lot of trial and error when developing for ArcGIS with python and it seems less then efficient to run things from start to finish to find out if some new code in the middle broke the process.

I have been reading about python's Unitest but I am having trouble applying it to Arcpy and geoprocessing.  

If anyone has any advice on test driven development with arcpy or geoprocessing in general it would be appreciated.