Another update in today's "research day"... Been reading an article/comment called JavaMail Reliability - Three Nines and Counting... from Brute Squad Labs. Though BSL has missed their own deadline for releasing their project "Iocaine", it seems to be a pretty fair and interesting read about how good JavaMail is, and how this fellow feels JavaMail does on the reliability scale.

This fellow loves unit testing, to say the least. I'm not sure I'd be happy that half my code for a project is for unit tests, but then I can't ever decide if I like eXtreme programming either. Code inevitably gets better the more [well-informed and skillful] eyes that pass over it, and gets better several orders of magnitude more quickly when other sets of fingers have to code using it. At the same time, it's easier (imo) to figure out the first gross or so bugs on your own. No reason two people have to sit through bone headed off-by-one errors. That's a waste of time. And before you say Person 1 will catch Person 2's obo error any faster than Person 2 wouold alone, s/he won't. When two people talk while coding, the appearence of obo [and other careless] errors tends to occur in stunning coincidences, like when pendulum clocks all tick the same when stored near each other.

Of course I realize there's more to XP than two people sitting at a desk, and I *think* I like the concept of unit testing, but I think I like to sit down and start coding towards the whole without bothering to make tests for each baby step and then refactor. But then I would; h'it don't make it smart.

For some more interesting reading, I got a kick out of their experiences with a Mac (they're testing their product on OS X) in their article Adventures in Multiple Operating Systems with Java, which really doesn't have much to do with Java.