Can I say "thank you" about his point on reusable code? Re-editable code as he puts it is far more useful.
Witness in puppet: their 'go to' module for example purposes is for ntp. Yet, last I looked, their ntp module contained a couple thousand lines of complex puppet logic to handle every configuration option under the sun. In making a meta configuration language, they have made a hydra of maintain-ability, in the name of re usability.
Far better in my view is a straightforward example that interested parties can take and tweak to their needs. Some need an iteratable list of ntp servers, some need a local clock skew.
Almost all ntp server environments can get by with a 20-30 line config file. Don't maintain a thousand+ line monster to maintain it in a reusable manner.
And now up the complexity to, say, tomcat. How complex would a fully flexible tomcat module have to be? Now do the same for Nginx. It's a pipe dream of combinatorial explosion.
Since most clients are supposed to work through a specified resolver rather than run their own, the easy block is to deny port 53 to non approved resolver hosts. Probably a good idea anyway in a secure environment, since it can potentially avoid cache poisoning if DNSSEC is setup right.
I call bs. I'm certain there will be a large scale problem with XP. However, that won't run in the reputation of MS, but rather the company still using it.
I'm typing now on a u6430 running ubuntu 13.04, and it runs great.
Three challenges:
1) On a busy wifi, I have some challenges with connection drops, but I haven't determined if that's a driver issue, or a chip issue. The connection drops are mostly after several minutes of idle network connections. (Now that I think about it, its probably driver, since rmmod/modprobe the wireless modules fixes things)
2) The thumbprint reader doesn't work out of the box, but honestly, I'd rather type a password.
3) When I try to use a usb3 'docking station', it works. However I find that I need to rmmod the usb3 module, and then modprobe it again to get it back to life after ~10 undock/redock operations. After about ~20-30 undock/redocks I find I need to reboot.
Dell makes a nice bit of coin (2+bil in profit last quarter according to google), mostly from the enterprise stuff, hence the kerfuffle over 3Par 2 years ago.
What I don't get is why they're submitting to a leveraged buyout.
(that said, I do see the utility, since it gives a more obvious visual queue as to the order of size differences... but if you're doing anything with the sizes programatically, you have to remove the commas afterwards... Short version: if you're going to do this, make it a unique flag, or a new flag modifier to the -l flag... don't overload the -l flag without recourse...)
GOW seems to be more focused on a userland environment, whereas mingw is focused on enabling developers to link against unix libraries in a windows environment. Sure there's msggrep, and msgfilter as userland in mingw, but GOW gives you direct access to 'ls', 'gfind', 'grep', etc, and doesn't care a fig about giving access to libz.
Personally, I think Gow is awesome. The few times that I need power unix stuff, it's been just there, and it just works. That's the most important thing.
For me, it's worth the 'price' of admission, just to get a usable instance of 'gfind', and avoiding the cognitive dissonance of typing 'ls' and not getting useful info.
People have mentioned syntax-highlighting, which is cool, and I'll be checking it out shortly.
But MY favorite feature of zsh is ZLE: the Zsh Line Editor. The fact that I hit up-arrow, and see the for|while|if statements as I typed them, rather than mashed together into a horrible, semi-colon infected, one-liner (like bash) is awesome.
Trust me. I get it. But if you don't have the HS diploma, it'll be something you have to explain for YEARS. People have certain expectations for others, even in high tech. The HS Diploma is one of those things. The college diploma still has a lot of the same stigma, though it's weakening. This stigma is a risk you'll have to weigh in your college or startup decision as well.
As for 'deeper in the stack', well, learning assembly in various guises is a good first step. Understanding how the pieces fit together, and how to make things faster is the next. I'm not sure if you're ready to start absorbing Knuth and the Art of Computer Programming (for reference, I'm not sure I am either), but that might be a good start too.
Oh, and another thing: understand the math. Learn the math behind this computer stuff. Understand how to measure the complexity and cost of a given algorithm. Something that might help there is Project Euler. It's something I enjoy hacking on when I have free time.
<-- hopes my children learn to code like you have.
I guess it depends on what else you're doing. If you're in High School still, suck it up and finish with high marks and be done.
As for college, there are two theories on that: on one hand, a college degree from a good school still caries a lot of weight with people, and it's a great networking tool. You'll meet some great people as well, and you might even learn something (though at your presumed skill level, you probably won't learn much of value until your junior/senior level CompSci classes. More valuable might be hitting CompEngineering or Electrical Engineering instead.
There's another theory which says 'skip college, do a startup', and there's lots of good reasons to take that route as well. It'll be less fun, more work, but possibly more lucrative, and you'll have much more 'working world' experience when you're done than your college friends.
As for 'what should you learn?' My MO for that is 'when in doubt, go deeper in the stack'. Those who truly understand the entire architecture of computers, and know how to drive the maximum performance out of it will always be in demand.
Also, work on the people skills. Learning to be a better communicator and how to connect with people will never be wasted. Negotiation is a skill to work on as well.
Well, Oracle is more prevalent in big enterprise. But if you look at the equations at the end of the article, a mysql/postgres DBA costs probably 80-90% of the Oracle DBA. The licensing costs are less, but the hardware cost for the big-iron and big-SAN to run the RDBMS is pretty much the same.
The big problem is in proving the collusion. It's just as easy, and mostly legal to send signals to your competitors through the marketplace... release a new model with foobaz feature at a higher pricepoint, see how the competitors react... if they add foobaz feature, and stay close to your higher pricepoint, you're safe, and legal. If they undercut, you can undercut to match.
Just look at the price delta for those TV's with Netflix streaming vs those without. You can't tell me it costs $300ish to add the parts to make this work. Wireless chips cost a few cents... the screens already have a CPU in them to handle everything, so add a few cents, or maybe $1 to add a better CPU...
I do really miss the 'command' module's convienience... so much that I often code my own version using subprocess, just so I can get at Std-out, Std-Err, and the return code all in one place.
Since both bash and zsh are 'sh' derived shells, it's not a huge context flip. It's analogous to flipping between your highly tuned (vim|emacs) config, vs logging into a new AWS image with the default version's config: You still know how to get around, but all the bells and whistles are turned off.
Witness in puppet: their 'go to' module for example purposes is for ntp. Yet, last I looked, their ntp module contained a couple thousand lines of complex puppet logic to handle every configuration option under the sun. In making a meta configuration language, they have made a hydra of maintain-ability, in the name of re usability.
Far better in my view is a straightforward example that interested parties can take and tweak to their needs. Some need an iteratable list of ntp servers, some need a local clock skew.
Almost all ntp server environments can get by with a 20-30 line config file. Don't maintain a thousand+ line monster to maintain it in a reusable manner.
And now up the complexity to, say, tomcat. How complex would a fully flexible tomcat module have to be? Now do the same for Nginx. It's a pipe dream of combinatorial explosion.