This may be sacrilegious in this crowd, but fear of patent suits is one of the major (perhaps *the* major) reasons that many companies don't open source more software. Device drivers are one of the most common areas where this problem crops up: if they open sourced their drivers, others would have lots of material to base a patent suit on. What others don't know about, they can't sue about. It sucks, but the system is what it is.

If you don't take the money, this is what will happen: you're clearly doing something they want. If you turn down your offer, they'll just have to reinvent it. They'll probably even be ethical and clean about it so you won't be able to sue for IP theft. But you'll end up as a few geeks with principles trying to compete with someone who has real resources. At that point, you're toast.

I strongly recommend "Feynman's Lost Lecture", a reconstruction of a lecture that Richard Feynman once gave that was a proof of Newton's equations as applied to planetary motion. All of Kepler's laws are derived during the course of the lecture. When Feynman prepared for this lecture, he set himself the challenge of doing it all without using advanced calculus, and restricting himself to "high school" mathematics. It's brilliant and totally do-able for (bright) high school students.

by Jon Bentley. It's easy to be fast.

I've had phenomenal success with OpenSolaris. ZFS is the coolest way to run a whitebox JBOD. The box I built can easily drive a GigE to saturation.