We finished our first XP iteration last week.
My major observation - watch your estimates. Because the iterations are relativly short, one bad estimate can really fuck you up. One of our stories was initially estimated at 20 bananas, but when we broke it down into tasks, we had nearly 300! Since the first iteration (Acorn) was bound to carry a bit of overhead in terms of infrastructure, we were only intending to schedule 400 bananas for the whole iteration. Doulbling our estimates almost straight away was always going to be a bit of a problem.
(A banana is a an arbitrary unit of time - not an hour, but hopefully not too far from being an hour, either. Establishing what a banana is actually worth - which will take an iteration or two - should give us our velocity.)
Anyway, with a bit of hard graft, we didn't do too badly in the end. We delivered just about everything that we were intending to, bar one missing major function and a few missing system table maintenance functions. Oh yes, and a few missing unit tests, too - bad Simon.
We demoed it to our directors (our internal clients) yesterday. It seemed to go pretty well - everything worked, and after a shakey first five minutes, I even made use of some of the presentation skills I'm supposed to have been taught. I think the directors were more impressed by what we've done than we are!
(In fact, one of the Big Four missed yesterday, so I'll be going through it all again this afternoon with the help of my glamourous assistant, err, I mean my assistant, Tulna.)
Next week, another planning meeting, and round we go again. I'll be keeping a more alert eye on all the estimates this time. ;-)
BTW, here's an XP practice that I'd like to see us put into effect - Beer O'Clock.
Posted to Software development by Simon Brunning at November 01, 2005 01:19 PM