On Fri, 09 Sep 2011 19:22:49 +0200, Thomas Schwinge <thomas@schwinge.name> wrote: > Ah, that's indeed a good approach for bug fixes (and it also preserves > git bisect compatibility), but still: why separate patches for new > functionality? (I'm not trying to be a pain here, but would like to > understand your rationale behind this.) I tend to think of the test as an actual spec of the behavior I'm trying to implement. By defining before hand exactly what I expect to happen I can confirm that my patch achieves what I want it to. As an example, you might look at my patch that adds better rfc822 part handling [0]. I tried to fully flesh out what I wanted to happen in the test first, and then modified the code to achieve that result. jamie. [0] id:"1307320169-29905-4-git-send-email-jrollins@finestructure.net"