I recently held a company-internal TDD workshop (which I hope soon will show up on Softhouse Education), and one of the questions that arose was whether there are any studies showing that TDD really works.
The feeling me, and most of those I've talked with, get by writing tests first is that it helps you chisel out the API on a quite low level, and that it keeps us from doing mistakes in the API design. This keeps defect rates down, and actually increases the speed of development, long-term. But I've yet to see a study with hard figures on this feeling.
There are studies made, of course, but all of the ones I've seen tend to lack some aspect. For instance, the paper On the Effectiveness of Test-first Approach to Programming only deal with students. I'd like the experimentations to be on a team with a mixed skill set, like normal projects. Another example is a TDD paper which was published very recently from Microsoft Research, Realizing quality improvement through test driven development: results and experiences of four industrial teams, which is super interesting. There are figures on the quality improvement (40-90% less defects). However, there are no hard figures on the change in development speed, which I'd love to have as well. The figures in the paper speak about a "15-35% increase in initial development time after adopting TDD". The problem is, these numbers are management estimates, numbers which naturally aren't proving anything, due to them being - estimates.
So, dear readers, have you stumbled upon any other research papers which deal with TDD and the points I've mentioned above?
0 kommentarer:
Post a Comment