Probably not very well. At that time in my life/learning curve, testing usually consisted of running that portion of the app that used the code being changed, trying various things that the users would do and calling it good. Sometimes I would combine that with logging the imputs/outputs and intermediate values while running the app and the sift through the results to see if it did what I expected. In time I was able to create a test harness for some parts, but realistically, that kind of testing only seemed to work well for those pieces/parts could be isolated from everything else. Very ad-hock. Part of the problem was that almost everything (business rules, data access, etc.), as delivered, was buried in the GUI :-(. | [reply] |