On Jan 9, 2019, at 2:04 PM, John Rose <[email protected]> wrote: > > you might consider using > a little combinatorial code to generate bad and good class > names
P.S. To motivate this suggestion a bit more: I found no problem with your manually-written test vectors of bad and good names, but I also found it difficult to be confident that they covered the ground adequately. Combinatorial logic for generating test vectors is often overkill, but it is easier to be confident about. That in turn makes us more confident that the actual code itself is free of bugs. Also, when I use such tactics in my own test development, I often find more bugs in my code, than with my first cut tests, which are small ad hoc test vectors. I won't promise the same for you, but it's a possibility.
