This was introduced with Java 1.6 update 10. If the compiler notices that the object isn't passed outside of a function (called escape analysis), it'll be allocated on its stack.

Mod Parent Up

The point is that a good tablet with more functionality than the iPad requires a good amount of research into how to do tablet UIs. The WIMP system is pretty terrible for tablet computing. That's why the iPad's an overgrown ipod touch, to avoid having to either do the research or be sucky.

Frankly, I'd love to see something designed for a stylus that also can take a few gestures usable for the hand holding that stylus.

It'd be nice to reduce the ping times between tokyo & dc.

Hey, if you know a better one, I'm listening.

A lot of people want good supported Unix servers with a bit more vendor-support (e.g. you want the guy who wrote the code to fix your kernel) than what most linux-server shops can provide.

Sun's had years with the right products, with some huge gaps, to do well here. But they've had their heads up their asses.

Or (now) Oracle DB threads.

Those bits you generally use STL for. But: (1) if an applicant can't do that, they usually can't do much else. (2) custom data structures are pretty common (and required).

Without that threshold, you end up with programmers who only know how to look up API documentation and call it.

I'm finishing my PhD now, while working. I just got hired a year ago, and make about as much mentioned (+/-, if you want to count guaranteed bonuses, etc.). Great benefits. The software developer market, for people who actually know what they're doing (e.g. C++, not PHP), is *hot*. Recruiters are calling everyone (even at work), and I'm going on my second recruiting trip next month. Anyone who can remember any specifics from the last 3 years of their undergrad CS degree would be nice. My employer hires non-CS and trains them how to program (for *months*, paid at full salary the entire time), if we can determine they're smart enough to learn.

The real issue is that most people calling themselves programmers can't even write a linked list or binary tree *TYPE*DECLARATION* without spending a half hour on google. They don't get hired, because they're not very good. But they're happy to complain that they don't need it in real life -- which is true, for the lower-paying jobs they'll get hired for.

God forbid vendors actually start testing their software *before* it's in the field.

IANAL, but after the 7min mark, is considered a crime against humanity in most civilized nations.

'based upon' is trueish, but in a funny way,

Mach was written out of BSD (modifying's easier than rewriting!), but the final kernel was clearly no longer a unix kernel. For Nextstep, the microkernel was recombined monolithically.

The research required for something better hasn't had funding for decades. Modern UNIX has been good 'nuff. It's got plenty of problems, but none big enough to justify a research budget big enough to rethink the OS.

Actually, it's only really been Sun pushing things forward recently, and it's mostly incremental.

Unix first. It was rewritten in C. http://en.wikipedia.org/wiki/Unix

Unix came out in '69, C in '72.

I've heard some pretty amazing government fraud stories. The best so far is a guy just making a bill in Excel and sending it to the Navy. They ended up paying $3 mil before catching him.

I've been looking at hp c3000 chassis office-size blade servers, which may serve as your production+backup+testing setup, and scale up moderately for what you need. Compact, easily manageable remotely, and if you're good about looking around, not terribly overpriced. Identical blades make a nice starting point for hosting identical VM images.