I got in at 6:11am to push the latest version of our e-commerce toolset to the live servers. It was dark and rainy outside. Everything worked which was nice.
When it comes to XP and future changes, I think the trick is to design something that will work "now", allow you to meet medium term goals, and just as importantly make sure that you won't design yourself into the corner in the long run.
A good example is what I did today. A couple of months ago we know we would probably need to support orders with items going to multiple addreses. We didn't worry about about it as we were concentrating on what we needed to implement to meet deadlines. This caused problems because we had to fiddle with both the code and the database because we had used the product id as keys, which isn't too helpful when you start wanting to be able to have the same product going to multiple people.