Well, either that or "every game I play" evaluates differently for him than for you. If I stuck to things that came out before 2000, then I could make the same claim about a Radeon R100 with a Pentium 3 CPU, too.

It would, although they'd probably taint it in peoples' minds, teaching it as a part of a mathematics course, or something. Looking back, a lot of my high school geometry course was similar to some of my theoretical computer science classes.

Slashdot is a minor player. The smallest TV news networks have larger audiences. Mainstream media and industry have done more to add a second definition (the one that you disagree with) to "Computer Science", in the minds of the public. Human language in general is stuck in a kind of out-of-control feedback loop, and that's one reason that we end up with islands of population-specific argot. It's useful for those groups to have more concrete definitions of words to use among themselves. I suppose that "Computer Science" acts as a kind of shibboleth; if you take it as a synonym for "Information Technology" instead of "The science of computability, computation, and information theory", then obviously you aren't a computer scientist.

Because it pays better than academia.

Programming is useful as a tool for studying computer science, so it tends to be something that CS grads learn, to varying degrees. The software development process can be learned on the job. Companies look for CS degrees to mean "software developer"; I don't see many job postings around me for someone who has a SW Dev degree or certification. It's less of a fight to just major in CS and learn to apply the theory to practical engineering problems than it would be to convince a company that they don't actually want a CS grad.

More like a biology major trying to get a job as a doctor, IMO.

I know two flavors of CS grads: Those who want to get jobs as academics and do teaching or research, and those looking to bend the purpose of a CS degree to get a programming job. In theory, it's a terrible fit. In practice, it's quite workable. After all, if one has the capacity to work out a clean mathematical proof, then one also has the capacity to build a clean piece of software.

I've never seen that, but I can't say that it's unbelievable. On my device though (admittedly, a much faster phone), it's always taken under five-ten minutes to ART-compile about 150 apps, and the idea that invoking each of those 150 apps one at a time would cause a similar amount of JIT compilation has always sounded reasonable to me. Since it seems to work in my case, I haven't really looked into it.

I've seen too many memory leaks to try to claim that it couldn't be improved upon.

It's better than JIT on Dalvik, which all non-Lollipop versions of Android would be using. You're complaining about compiling once, when the program is first installed. How about compiling once per time that you run the program? ART provides a net benefit over that scheme.

The "source code" is Java/Dalvik bytecode. ART just replaces the JIT compilation that used to run every time the code was executed on Dalvik with a one-time compilation to native code. Is the bytecode actually that much smaller than an X86/ARM binary?

Ah, yes. The OS that notifies my wife to reboot her phone daily, to avoid memory leak problems that haven't been fixed since she bought it. The OS that requires her to convince herself that "No, I'm not as interested in that app as I thought", because it's not available on the app store and has no alternatives available. The OS that goes into reboot loops if her phone goes out of range of a cell tower. Yes, I've used Windows Phone. If it worked properly more often and had a better app ecosystem, it would be a perfectly useable phone OS.

The version of mathematics where there's a $200 version of the phone, and they specified a $700 phone as the comparison. At least some versions of the Galaxy S6 are $700 when purchased off-contract, and the iPhone 6 seems to be around that price too.

My manager has told me "Software development isn't rocket science. I don't see why investigating a bug takes so long." She also hasn't done any development work herself, despite having a computer science degree. She went almost directly into a PM role out of college and doesn't have a frame of reference to understand her employees' work. She hasn't ever been able to understand what her employees are doing, what's reasonable in terms of timeframe for a feature or bug to be completed, doesn't know the product well enough to even know which part of it her team members are working on, etc.

In contrast, my previous manager was promoted up from the ranks of the developers. She was amazing to work for, because she always made her team's jobs easier, and understood what we were working on (in a broad sense, not down to the deep details, of course).

Knowing their jobs isn't the same thing as taking the time to do their jobs. A manager should manage, but they should know what their decisions mean to their team members. To my current manager, a flurry of e-mails means work is being done, and she expects a constant flow of e-mail to go around her team. I've been chastised in her office several times for not communicating enough while working through bugs on a piece of code that no one else on the team has experience with. I explain to her which bugs I've fixed, which I am still working on, and which features I've implemented. Doesn't matter. No e-mail means no work, because that's how her world works: E-mail and meetings mean decisions are being made and work is being done. No e-mail or meetings == black hole where nothing is happening.

Ditto. I know my work better than my manager. That's basically by definition, except for in a very upside-down project. But if I say "I've got the fixes for these 3 bugs in, and I'm working on these 4 others", and I get a reply like "OK, but what are you actually doing?", you'll have a hard time convincing me that I'm not working for an idiot. They don't know my job, don't understand what their employees are doing (even in abstract), and don't belong in a management position over a team until they understand what it is that they're managing.

If you're over 40 and you don't know how to answer that question based on past experience, I think you're in trouble. Picking up new languages, frameworks, APIs, and what have you are just par for the course. Those things have been a constant in every development job I've had. If a language is related to something that I already know, then within a few weeks, I may be writing some Perl-ish looking Python and becoming more comfortable using constructs that don't appear in Perl very quickly.

If math were taught correctly, it would have a lot more in common with a study of structured reasoning and logic, rather than memorizing algorithms for long division and lookup tables for arithmetic. Math education shouldn't be abandoned, just fixed.

Parts of that wouldn't be feasible (notes coming in about sports scores, news items, what have you), but keeping a local database to correlate locations, time, and all that? Sure, you could do that. Ditto with the voice recognition, especially if it's limited to a certain set of vocabulary that's useful for on-device management of your alarms/calendar/contacts.

I wouldn't? That's what I've got a Garmin with radio traffic updates for. I bought it before I had a smartphone, and it still works as well as it did when I got it.

Oh? I can't host an AP from my laptop, run a VNC or RDP server there, and connect from my tablet? Or run VNC on my tablet and access it from the laptop? I'm not sure what you're claiming, because the most direct interpretations of your post have counterexamples.

Well, useless for anything that needs internet, sure. I can't check email or browse a webpage, but I can program, write, play games, listen to music, watch movies, etc. I mean, it's not like you're using Steam for all your games, Spotify for your music, and Netflix for your video, right?

SDL and SFML are both pretty cool and have large development communities. I haven't used Allegro or Clanlib, but Clanlib's features seem especially close to SFML's (based on examining the API). Allegro has been around for over 20 years. SDL for 17, Clanlib for 16, SFML for just about half of that.

It raises a question in my mind: If Clanlib had been out for that long in 2007, providing a C++ game programming library and being well-known enough to be included in at least one book, then why was there a perceived need for SFML? Was it just a marketing/popularizing failure on the part of clanlib's developers, or was there some technical shortcoming in Clanlib at that time? If the latter, has the shortcoming been corrected? In short, why exactly should we be excited about a new release of Clanlib?

They give me more, and get paid much more. My Facebook has about as much information on it as one of my old instant messaging accounts did. It gives you rope to hang yourself with. Doesn't mean you have to use it.