On threads such as this one it is easy --and even popular-- to dump on those, such as myself, who come down on the "nay" side, even when this is firmly based on

On threads such as this one it is easy --and even popular-- to dump on those, such as myself, who come down on the "nay" side, even when this is firmly based on experience and having made plenty of mistakes in the field. The linked article does NOT talk about a one-semester school project or a quick-and-simple learning OS. No, the article talks about a web developer with no real experience writing low-level code not only wanting to bootstrap every single device driver but also ignoring years of accumulated knowledge and code libraries to write an OS that boots directly into graphical mode, does not take advantage of POSIX and more. There's nothing wrong with the "How hard can it be?" approach to learning. I've done this many times. And almost every single time I came away with "I sure learned a lot, but what the fuck was I thinking?". The last time I pulled one of those was about fifteen years ago and the "three month project" took nearly two years. What he is talking about is more complex than writing the Linux kernel from scratch because he wants to re-invent everything. Here are some stats on the Linux kernel: http://royal.pingdom.com/2012/04/16/linux-kernel-development-numbers/ http://royal.pingdom.com/2012/04/16/linux-kernel-development... Even if his project was 10% of this it would still be a grotesque miscalculation for a single developer, otherwise employed and without the experience to back-up some of what he is proposing. If, on the other hand, the post had suggested something like this it would have been far more reasonable an idea: "Hey, I just spent a year implementing everything in the Tanenbaum book. Now I would like to start from that base and enhance the OS to make it do this...". Let's compare notes in a year and see how far he got.