Re:Logic (Score 1)

I managed a project that piloted a few parts of XP and were extremly happy with the results.

Testing and Pairing was a big thing for us, and certain interesting things came out of it:

1) a couple of the developers that were most opposed to pairing turned out to be legends in their own imaginations, and their reluctance was justified when they were exposed to other people
2) those pairing must burn into it. Yes, you have to sit back and sometimes get to know someone, but that is the benefit. It often took a couple of days before developers were comfortable with each other, but all admitted they were way more focussed when paired than alone. I made sure everyone could kick back and take an email/surfing break though - pairing is more tiring - those that paired usually tried to do their best work all the time.
3) We had all developers take a hand in the new coding standards before we started. Hence, there were very few stylistic bickerings or religeous wars about brackets.
4) Judicious pairing benefits the pair very well - a good combination was good OO and C++ in one head and good C and UNIX system development in another. Both parties learnt from each other, and produced a tight library that was bug free and on time
5) Testing up front really helped refine the requirements.
6) Confidence in the shipped code was very high. There were no bugs, and due to using testing to refine the requirements early on, the library was accurate to the spec.

Our initial foray turned out quite well, but we have other problems - a rather major legacy code base that we can't ignore. We're looking at ways of rolling XP into some of the development here. We accept that XP is currently good for green-field work.

Reading this thread, I also noticed that the detractors seldom seemed to have read the book, dismissing it as 'yet another methodology'. :-)

M