The only way to justify owning this ridiculous amount of white plastic would be for each peripheral to be supported by a pretty solid library of games. Instead, we see a mere handful of low-quality games making halfhearted use of them.
Have you never owned a Nintendo device before? When was this ever not the case? Have you noticed that Nintendo never sells their consoles at a loss (unlike Microsoft and Sony, which have a reputation for it)? Same as with Apple, the money is with the hardware.
if you have 30 yrs on your resume and most of your jobs had programming in them... you don't need to insult him with syntax questions and all that.
My focus with someone with that level of experience isn't basic syntax. Syntax is easy. I've seen plenty of older developers who still retain stale prejudices from back in their heyday. My concern is getting someone who thinks age = knowledge and closes his mind to learning something from a young whippersnapper like me. I've seen at least one candidate who was obviously offended by my presence in the interviewing room (he wasn't hired).
I'm on the other side of that fence. I've had to spend a fair amount of time reviewing code submitted by candidates. Trust me, it is burning up the company's time as well (well, at least in my case...), and it does provide a lot of additional insight into potential candidates, if done well.
I've never had anyone point me towards open-source contributions, but that would probably be a very big plus and potentially waive the "take-home problem". I'm sure we've lost a fair amount of good candidates who share your sentiment—but we've also weeded out enough poor candidates that it's worthwhile overall.
It's a major character flaw of mine to fail to apply ESP and give people what they want rather than what they ask for - I'm working on it....
I know a guy who argues the same point (nobody sympathizes with him either). Kinda makes it hard to deal with any social aspect of humanity, since most communication is nonverbal or implied.
"buffer overflow" != "coding standard preference". If something's wrong, it's wrong, no style guidelines needed.
Though I think it's a bit harsh to knock someone out because of a single error. What if that was the only mistake? What's the likelihood of a reviewer seeing a "perfect" solution that has 2 overflow vulnerabilities? Better to read through the whole thing and respond with a more comprehensive list of problems, and see if they get back with patches / updates to their code. It's more time-consuming, but it's worth it to select the right people to work alongside.
the next programming test I take, I'm insisting I bring a laptop, have emacs and gcc at my disposal. I mean, I do NOT write code on whiteboards with markers in my real job, why should I have to put up with that in an interview?
I'd refuse your request. Why? Because when I'm asking for someone to provide a white-board solution, my expectations are completely different. Here's what I care about:
Do they have a passable-enough knowledge of their toolchain to not require looking up trivial stuff? The compiler and google help with a lot, but it sure hurts when you're reaching for a reference for every line of code.
Can they reason through an algorithm? Code is worthless if it's not planned, and planning doesn't require syntax highlighting.
Can they think on their feet? I don't want to spend 10 minutes getting a solution that compiles without warnings, I want to get the conceptual problem solved in 2 minutes and spend the next 8 minutes discussing the merits of the solution, possible bugs, alternatives, and any rationale/motivation for the solution as presented.
As to whiteboarding on the job... if a co-worker can't express his idea of a solution without resorting to emacs + gcc, there's a good chance that the idea either isn't good to begin with, or that it will be poorly implemented/maintained due to a lack of shared understanding. Good design leads to good code, not the other way around.
Other posters commented on how vim is well-integrated with make and other development tools. Granted, the toolchains may be separate, they do fit together. With Visual Studio, the build toolchain is packaged along with the editor. But if you look under the hood, it's not that difficult to split apart the toolchain and make Visual Studio work in a foreign environment. The only major difference at that point is that Visual Studio comes with a GUI for configuring your compiler/linker settings, and vim doesn't... and that's more of a cultural difference than anything else.
Re:C# isn't a language...
on
Head First C#
·
· Score: 0
The main problem in C++ and C# is that just looking at a statement you can never be sure what it does without poking around a lot of places (in Java a statement always says exactly what it does). A simple statement like "x.y = a[b]" can involve at least 6 different method calls. In Java is always means assign array element to a field.
In theory, you are correct. C++ allows developers to shoot themselves in the foot in so many "interesting" ways. In practice, if you don't know what a statement does just by looking at it, you locate the original developer and give them a solid WHACK on the head for writing obtuse interfaces.
And that's why you should only run as admin when you need to do admin stuff. Do all your day-to-day work as a Limited User.
OK, so the operating system won't be hosed, but all his user data will be? That doesn't sound too helpful, given that the operating system can be trivially recreated; all of the data in "My Documents/Pictures/Crap" is still accessible (deletable) to a virus that runs as said limited user.
Well, cross integration of command line tools is acceptable (pipes etc), but they missed the boat a bit with gui based apps...
Fixed that for you. I'm a fan of the UNIX way of combining applications to make a "solution", but the standard approach is very limited. If you're transforming data and it's always representable as rowsets, piping works fine... but for advanced tasks, I always end up with a mixture of bash, awk or even XSLT processing. Sometimes I want more than just a sequence of streaming text processers, and IME there isn't always a consistent way to get that between processes.
Incidentally, if you haven't already seen it, check out DBUS and DCOP. Those are both attempts at providing hierarchical attribute access and control (think "Firefox config") between processes. Not quite enough to replace the stdin/stdout paradigm, but it's a step in the right direction. Treat all of your GUI apps the same as network peers, and then the distinction between "desktop" and "cluster" get blurred real fast:)
You can get the same effect as two eyes by rapidly shifting focus from far way to close and analyzing the differences between them to gauge depth.
It's not quite the same effect. I'm assuming you mean looking for sharp edges and observing how their "blur radii" respond to changes in the focal distance. It still lacks true depth perception, and I can't imagine that working well for a smoothly-curved surface like a face.
habitually inhaling toxic smoke... contrary to any living creature's survival instinct
Whoops, you just made the assumption that survival instincts are based off of long-term cost/benefit ratio. On the contrary, most instincts (fight-or-flight, sexual attraction, whatnot) are based off of estimated short-term benefits. Hence why people continue to smoke...
it satisfies an immediate craving (if you're already addicted)
nicotine gives a nice little buzz
it is perceived to reduce stress
Regardless of its truth and long-term effect, the "inhaling toxic smoke" argument borders on propoganda. It completely ignores one side of the argument.
The JIT compiler is very impressive - it keeps basic profiling data when running the code as bytecode & compiles to native after 1000 executions of some code...
Is it possible to store profiling information between runs? That's the big difference I see between in-house C++ and Java. Sure, after 1000 runs of any given function it may be optimal (and better than the equivalent statically-optimized C++)... but what about before then? For real speed, I find myself compiling C++ code with profiling enabled... then running a real-world load test... then recompiling with profile feedback, which can substantially improve code generation.
If the Java JIT could take optimizations from one run and apply them to the next, that would be awesome. But if it can't, then it's worthless (to me), because it means that restarting a latency-sensitive application means "forgetting" all of its optimizations.
I get what you're saying here... gg=G. I think this is just an issue of a lack of automatic indent-on-paste in vim. It would be very reasonable for vim to recognize "hey, I know this is Python code, so when I copy-and-paste, I should adjust the minimum indent-level to match the current indent level". I'm no vim-guru, but that makes more sense than forcing a re-indent after the paste action. In fact, I'd go so far as to say that the braces-style reindent is only required because indenting can't be inferred on paste in a language like C/C++/Java/Javascript/...
I don't think it sounds crotchety to use vi... I'm mostly fresh out of college, and vi is my editor of choice... FOR PYTHON! Seriously though, your approach to python sounds like my approach to swinger parties and wife-swaps. Deathly opposed, but probably only because of infamiliarity.
I choose "Pearl", because it's not a word I commonly hear among developers. Sounds a lot like "Perl". I would say "it's not a programming language", but I know someone will come out of the woodwork and cite some obscure language named "Pearl" that nobody cares about.
Slashdot Mirror
I think you mean int main.
That's great!
Have you never owned a Nintendo device before? When was this ever not the case? Have you noticed that Nintendo never sells their consoles at a loss (unlike Microsoft and Sony, which have a reputation for it)? Same as with Apple, the money is with the hardware.
You could always Wii-Punch them in the mouth whenever they skimp on the motions. And you don't even need a Wiimote for that true facepunch action!
Fair deal.
My focus with someone with that level of experience isn't basic syntax. Syntax is easy. I've seen plenty of older developers who still retain stale prejudices from back in their heyday. My concern is getting someone who thinks age = knowledge and closes his mind to learning something from a young whippersnapper like me. I've seen at least one candidate who was obviously offended by my presence in the interviewing room (he wasn't hired).
I'm on the other side of that fence. I've had to spend a fair amount of time reviewing code submitted by candidates. Trust me, it is burning up the company's time as well (well, at least in my case...), and it does provide a lot of additional insight into potential candidates, if done well.
I've never had anyone point me towards open-source contributions, but that would probably be a very big plus and potentially waive the "take-home problem". I'm sure we've lost a fair amount of good candidates who share your sentiment—but we've also weeded out enough poor candidates that it's worthwhile overall.
I know a guy who argues the same point (nobody sympathizes with him either). Kinda makes it hard to deal with any social aspect of humanity, since most communication is nonverbal or implied.
"buffer overflow" != "coding standard preference". If something's wrong, it's wrong, no style guidelines needed.
Though I think it's a bit harsh to knock someone out because of a single error. What if that was the only mistake? What's the likelihood of a reviewer seeing a "perfect" solution that has 2 overflow vulnerabilities? Better to read through the whole thing and respond with a more comprehensive list of problems, and see if they get back with patches / updates to their code. It's more time-consuming, but it's worth it to select the right people to work alongside.
I'd refuse your request. Why? Because when I'm asking for someone to provide a white-board solution, my expectations are completely different. Here's what I care about:
As to whiteboarding on the job... if a co-worker can't express his idea of a solution without resorting to emacs + gcc, there's a good chance that the idea either isn't good to begin with, or that it will be poorly implemented/maintained due to a lack of shared understanding. Good design leads to good code, not the other way around.
Can't offer an NDA to show code that belongs to someone else. That's the situation most people are in - work for hire.
Anyone who claims proficiency with C++ in less than a month is either lying or a prodigy. Proficiency means much more than learning the syntax.
I got that for my dad!
Other posters commented on how vim is well-integrated with make and other development tools. Granted, the toolchains may be separate, they do fit together. With Visual Studio, the build toolchain is packaged along with the editor. But if you look under the hood, it's not that difficult to split apart the toolchain and make Visual Studio work in a foreign environment. The only major difference at that point is that Visual Studio comes with a GUI for configuring your compiler/linker settings, and vim doesn't... and that's more of a cultural difference than anything else.
What about vim makes you think it's not an IDE?
In theory, you are correct. C++ allows developers to shoot themselves in the foot in so many "interesting" ways. In practice, if you don't know what a statement does just by looking at it, you locate the original developer and give them a solid WHACK on the head for writing obtuse interfaces.
OK, so the operating system won't be hosed, but all his user data will be? That doesn't sound too helpful, given that the operating system can be trivially recreated; all of the data in "My Documents/Pictures/Crap" is still accessible (deletable) to a virus that runs as said limited user.
Creditworthy?
Fixed that for you. I'm a fan of the UNIX way of combining applications to make a "solution", but the standard approach is very limited. If you're transforming data and it's always representable as rowsets, piping works fine... but for advanced tasks, I always end up with a mixture of bash, awk or even XSLT processing. Sometimes I want more than just a sequence of streaming text processers, and IME there isn't always a consistent way to get that between processes.
Incidentally, if you haven't already seen it, check out DBUS and DCOP. Those are both attempts at providing hierarchical attribute access and control (think "Firefox config") between processes. Not quite enough to replace the stdin/stdout paradigm, but it's a step in the right direction. Treat all of your GUI apps the same as network peers, and then the distinction between "desktop" and "cluster" get blurred real fast :)
It's not quite the same effect. I'm assuming you mean looking for sharp edges and observing how their "blur radii" respond to changes in the focal distance. It still lacks true depth perception, and I can't imagine that working well for a smoothly-curved surface like a face.
Whoops, you just made the assumption that survival instincts are based off of long-term cost/benefit ratio. On the contrary, most instincts (fight-or-flight, sexual attraction, whatnot) are based off of estimated short-term benefits. Hence why people continue to smoke...
Regardless of its truth and long-term effect, the "inhaling toxic smoke" argument borders on propoganda. It completely ignores one side of the argument.
What do namespaces have to do with object models? I've never seen an ad-hominem attack against a programming language before!
Is it possible to store profiling information between runs? That's the big difference I see between in-house C++ and Java. Sure, after 1000 runs of any given function it may be optimal (and better than the equivalent statically-optimized C++)... but what about before then? For real speed, I find myself compiling C++ code with profiling enabled... then running a real-world load test... then recompiling with profile feedback, which can substantially improve code generation.
If the Java JIT could take optimizations from one run and apply them to the next, that would be awesome. But if it can't, then it's worthless (to me), because it means that restarting a latency-sensitive application means "forgetting" all of its optimizations.
I get what you're saying here... gg=G. I think this is just an issue of a lack of automatic indent-on-paste in vim. It would be very reasonable for vim to recognize "hey, I know this is Python code, so when I copy-and-paste, I should adjust the minimum indent-level to match the current indent level". I'm no vim-guru, but that makes more sense than forcing a re-indent after the paste action. In fact, I'd go so far as to say that the braces-style reindent is only required because indenting can't be inferred on paste in a language like C/C++/Java/Javascript/...
I don't think it sounds crotchety to use vi... I'm mostly fresh out of college, and vi is my editor of choice... FOR PYTHON! Seriously though, your approach to python sounds like my approach to swinger parties and wife-swaps. Deathly opposed, but probably only because of infamiliarity.
Come one... join us...
I choose "Pearl", because it's not a word I commonly hear among developers. Sounds a lot like "Perl". I would say "it's not a programming language", but I know someone will come out of the woodwork and cite some obscure language named "Pearl" that nobody cares about.