It's really a silly argument. The BSD license does have fewer restrictions, but that doesn't make it better than the other. I think people need to understand that the two licenses have different goals in mind, and developers need to respect the wishes of the rights holder. Likewise, developers should take care in what license they use.

My guess is that the BSD license's intent was to simply give credit where credit was due and to allow researchers to develop code for anyone to use, in proprietary or open source projects, with limited liability. This is a good license to choose if you want to give your code away and only want recognition.

The GPL's intent should be obvious to everyone here: The FSF is after a system entirely composed of open source software, and the GPL is one of their tools to achieve it. If you do not want to be a part of this community, do not license your software as GPL and do not expect to be able to use someone else's GPL code (in your own code). If you don't like it, tough--you may as well be complaining to a car salesman about your car not being free.

But if you hate the GPL and FSF, you might not want to use the BSD license. They can use your code too. ;)

Competent admins? You've got competent admins?! Lucky... It was a few weeks ago that we let ours know it's a bad idea to make home directories for class accounts world-readable...

I'm not a big fan of computer labs--I prefer the comforts and distractions of my apartment. Sometimes it's nice to hang out with your classmates, though, while you're waiting for that simulation or synthesis to finish. Or when the damn wireless chip doesn't follow the spec and you need to make sure you haven't gone insane...

But yeah, the expensive apps could be a problem. Using an app in the lab or buying your own copy aren't necessarily the only options, though. I wish for more floating license setups, where you run the program on your machine but tunnel through a login server to check out a license.

VLIW and out-of-order superscalars attempt to extract more performance out of the same type of parallelism, actually (instruction-level parallelism). The way I see it, they are just different approaches to the same source of speedup; VLIW just pushes a big chunk of the work from the hardware to the compiler.

Hm, I disagree. Science does not start from evidence. It begins from the irrational, from the formation of the models and language used to describe an observer's sensations. Evidence is used only in a model's falsification, made possible by science's crucial premise--the existence of a predictive, deterministic model. That is a belief, by the way, not something that can be proven (neglecting omniscience).

So it's not that she's discarding evidence or reason. She just may not share the same basic beliefs. Neither side is provably correct; moreover, that foundation confines both your knowledge and hers.

That said, without some of us believing in predictability, I wouldn't have a job! ;-)

Not all sites are like that, though. For example, HardOCP attempts to find the highest settings a card can handle while still achieving some minimum average frame rate. You still can't take the numbers as indicators for your particular system, though, unless all of your other components have comparable performance to their test rig.