Berlin Extreme Hour

On the 12th of October, 2001 members of the BerlinXpUsersGroup met to perform an ExtremeHour. A total of ten people meant that all roles were filled and everybody had something to do. We followed the structure outlined in ExtremeHour, with a couple of changes.

As product, we choose the automatic diaper changer, this meant that parents were stakeholders and quality assurance, and the rest were developers. In the end we had 4 developers, 4 stakeholders + QA. We merged the stakeholders and QA into one group, we felt that since XP does not have explicit QA people, it didn't make much since to have them.

The first ten minutes stakeholders wrote up a bunch of user stories on A5 cue cards. In the meantime, the developers worked on architectural spikes to get a feel for what they had to implement/draw. The groups were not physcially separated, which meant that the stakeholders were passing their user stories over to the developers. At the end of the iteration, the architectural spike drawings were thrown out.

User story Estimations followed in the next ten minutes. What we did was hang up the stories on a whiteboard, and developers and stakeholders huddled together to make estimation and prioritize the stories. In the end, we had just enough user stories to cover the 13.3 mins ideal time that the developers had for iteration 1. This meant that the third ten minutes where left for the developers to sort out who did what.

Iteration One was done with all groups in the same room. This meant that a lot of problems that would have come out in the functional tests, were already solved by the developers by asking for clarification from the stakeholders. The developers worked in pairs but did not rotate, and at the end of the iteration, unit tests were done.

After the stakeholders tested the results of the first iteration, the developers completed 10 ideal minutes of user stories, which then became the time allocated for the second iteration. An interesting bug that came out the functional tests was whether the baby actually survived the process or whether the changer needed to guarantee the welfare of the baby. In the end, because it wasn't a user story, it was decided that the developers did not need to have a check. This requirement was then the first user story for the second iteration!

In the second iteration, one developer pair left for another room while the other stayed put. But this time, there was less discussion between developers and stakeholders and we kept the developers busy by rotating pairs every 3 minutes. This also ensured that unit tests were carried regularly, and also helped to ensure that all the developers understood what was being built.

The end result was the developers managed to pass all functional tests and the stakeholders where help with the end product.

Notes:

• we went over time in every ten minutes bar the first two, the reason for this was the discussion both related to the user stories and the XP process.
• we photocopied the results of the first iteration (increasing the size from A5 to A4) and allowed the developers to continue in the second iteration using the copies.
• we didn't do the entire process non-stop, we had 5 minute breaks and in the end it took us circa 1.5 hours to do the entire hour.

All-in-all, amongst the group opinions differed on the effectiveness of the hour. For some it was clear how XP differs from other processes, for others the hour gave them a better overview of XP.

---

I've now onlined the cards we drew, check them out at http://munch.wiwi.hu-berlin.de/servlets/WikiServlet?BerlinExtremeHour=1

-- GerritRiessen