You ask them to play something that's easily gamed, you get good game players :/ Garbage in - garbage out.

It all depends. I know that if you call any of the GDI's filled polygon/filled shape APIs, it's all at the very least tessellated in software, and then perhaps if the hardware is quick enough to render a bunch of triangles as an instantaneous sort of a thing, it will pass it on, but I doubt that this has been done for any recent hardware. It requires very low overhead in starting a "job" on the hardware and is mostly suited for old style of graphics hardware that can fill a list of triangles without invoking the entire 3D pipeline (if it has any). It used to be a thing in the late 90s and early 00s. For 99%+ of common display architectures out there now, GDI is not really accelerated outside of blitting stuff.

If you use any graphics library that doesn't depend on GDI, you can probably do better even in software rasterization since you have enough knowledge to split the job and parallelize it. Or you simply pass it as a draw list to DirectX and let that do it orders of magnitude faster - but again, GDI doesn't do it since its stateful architecture is really a poor fit to modern asynchronous rendering pipelines.

Leap Motion's C++ API is exemplary. They must do something right to produce code that good.

Whereas in real life you'll often have to do things like this where first principles do matter and where you have to understand not only what the standard library is doing, but what are the side effects of the way it's doing it.

Yeah: Is it really so much to ask an experienced developer to prove that they can do code reviews? And if someone can't review code without an IDE, they're handicapped. The question remains whether their other qualities make such a handicap worthwhile. Perhaps, as a means of self-development, everyone should spend some time on StackOverflow and Code Review and learn to spot mistakes in what others do - iff they don't do code reviews in their current job.

Yep - there must be some screening, you can't pretend to be a software developer who can't write some trivial code correctly in any of the languages they purport to know. Hello World is a good first step - if you can't do that, you have no place pretending to be a developer.

Would you hire a pianist that can't point to a middle C on a standard piano keyboard? Because that's the equivalent of what's asked in the interview you speak of. At that point you don't really care if perhaps they can play all the scales very well - you can't trust the completeness of the internal framework of knowledge they use. People who don't truly understand what they're doing tend to divide and compartmentalize various areas of their knowledge instead of synthesizing a unified and interconnected body of knowledge. They learn all sorts of recipes to do various things, but have no idea how all of those things are interconnected, and what sort of mental framework unify them. It's as if a baker didn't realize that the yeast flatbread dough they deal with shares some common properties with the yeast-raised sheet fruitcake.

I think it's a particular state of mind where you expect to be able to forget the basics. I personally don't find this state of mind to be all that appealing. It's like if a concert pianist forgot all of their music theory, because for performance you don't really care about it. But then it kinda sucks if you're a great concert pianist who superbly plays very technically demanding music, yet is unable to harmonize a simple melody during a show-and-tell with some kids. Yes, it's great that you can put design and implement a scalable architecture for a big system of some sort, but it kinda sucks if you can't do the basics. It's as if an electrical engineer forgot an inductor's constitutive equation, having an excuse that they deal with fancy control systems all the time and haven't used any inductors in ages. It smells of functional illiteracy to me. Or at least I try to keep my basics refreshed to some extent, as a conscious effort where I spend an hour or two every week re-reading the fundamentals just to keep them fresh.

Yes, but only if the spec is for it to output the entries. Perhaps it shouldn't. Assuming the "obvious" can be pretty wrong.

Joel Spolsky had this to say about the chairs, and I agree with him:

You've hit on a key point: self-defense is acceptable and necessary, it's a basic human right. That right shall not be postponed and delegated to a third party - as is the case with death penalty.

Huh? I use the latest LibreOffice regularly on a late 2008 MacBook Pro with a 128GB Crucial SSD, on El Capitan. Works great, and that's a Core 2 Duo CPU with 8GB of DDR3 RAM. It'd be considered obsolete by pretty much anyone these days - yet it performs admirably.

OS X versions past 10.7 suck donkey balls on mechanical hard drives for some reason. The CPU on your Mac Mini has nothing much to do with its sluggishness. Replace the drive with an SSD and you'll fell like you've got a completely different, new machine. Just do it, you'll thank me later.

I do precisely that: each invocation of rm creates a fresh timestamped folder under mountpoint/rm_saved

God no. We moved off of Zimbra a few months ago, and Zimbra was already way easier to deal with than a standalone Postfix/Dovecot server would be. We're in the 50 user territory and O365 is a steal, even though we pay around $600/month for it.

Huh? You go to the admin center and can clearly see all of the plans and can choose accordingly. Choice is good here, not bad.