RE: [Dunit-interest] RE: XP Experiment

[Jon B. <Jo...@ci...>]

> But from what I have seen, the benefits of test-intensive development
> are substantial and obviously beneficial to a wide range of projects.

I agree completely.

As a side note.  I'm running my latest project as an experiment:=20

Call it "Minimal design.  Does it really work?"

With stories in hand I started writing test-first.  I watched the number =
of methods that needed refactoring (i.e. after the test passes I find I =
have to change the test because the required functionality changed on my =
way to "first release.")  (This is a program only for me so there is no =
formal release).  Two out of 21 needed refactoring.  Program worked with =
no trouble.

For the second release I continued on.  Major database change to isolate =
the db and make future changes easy.  I had a major design oops.  20 out =
of 42 methods needed rework to fix the oops.  Bother, it took a ton more =
time to get to second release. =20

I got nervous - maybe my version of "minimal design, test-first and =
refactor like mad" is a problem.  So I sat down and planned release =
three (major UI change to allow making changes easier).  Seven hours of =
design (it's a fairly small project).  26 out of 62 methods needed =
rework.  Same general type of oops happened.  Less code added but about =
the same amount of code didn't add up when I was done.

The code now seems "up the design curve."  It'll be interesting to see =
the next stats for the next version.

Unscientitic as it is I still use this kind of experience to help me =
form my view of test-first.


Jon B.