Last weekend I spoke at the Great Lakes Software Symposium in Chicago and an audience member verbalized a great tip about driving the team towards small code commits.
When you want a team a to add one feature or fix one bug, you use tools like continuous integration to train yourself (and your team) to check in a single feature. But how big are those features? If you've got features lasting weeks or months, then adding a single feature is a huge amount of code.
The key to small code commits is small estimates.
My definition of small is between half a day and a week's worth of work. Anything longer than a week probably has a surprise or two hiding in it. You haven't decomposed that feature down far enough and when you encounter this hidden surprise it might blow up in your face. Anything shorte
Previous Topic
