Well, the F22 was built for the same role as the F16 had initially: pure air-to-air combat, but now with supercruise to intercept at a distance. Nothing ever built even comes close. It's a damn shame we built so few of them, as the F35 can't hold a candle to it in that role, but with its operational issues it's understandable that it landed on the chopping block. (Seriously? Computers crashed when crossing the international date line? That's just embarrassing.)
We can't have multiple different aircraft when a cheaper alternative is one aircraft that does the jobs to almost the same level.
Well, that word "almost" is disputed by many. Plus it creates all the normal mono-culture risks: if it turns out to have some exploitable weakness, than all our planes will have that weakness. But given we won't be funding the military at "superpower" levels much longer (arguably we already don't), it will have to do.
Ultimately, the transition to combat drones is well underway, and I don't think it will end up being that big a deal if the F35 is second-rate. It's sort of the stopgap while we phase out piloted combat aircraft over the next 20 years or so.
"Horizon" is a fuzzy word. The APG-71 radar system, now deployed for 30 years, allows locking on to multiple targets at 230 miles as deployed (twice that as designed, but the antenna actually used isn't ideal). Pick what terminology you want, but I call that "long range".
The replacement for the F16 is the F22 (which sadly we have far to few of). The F16 was designed as an interceptor, a very different role than an air superiority fighter (and one that has become almost meaningless), but it works well enough. The F35 isn't in the same league as the F22 for clearing the skies, but since we're cutting every military budget everywhere, we can't have nice things.
But by that analogy, there are any number of MOBAs, for example, so there are several choices of publishers. Sure, for a given exact rules set, there's only one, but again that's the same as pro sports.
Think of the NFL as "football's publisher" and you have the same situation - monopoly control over professional gameplay hasn't been a problem thus far. The difference, think, is that congresscritters care about the NFL, and won't let it go off the rails too far (and the government has poked its nose in from time to time). It will be interesting to see how that plays out with eSports - whether something like FIFA will come to be.
All that sounds right to me. However, F-35s are already visible enough to high-tech, high-power radar for situational awareness - they aren't B2-style stealth aircraft - if only marginally so. Going to longer-wave scanning radar seems to me to be the same as just throwing out 10x the radar energy to get the same result - still important, but not a difference in kind.
Given we usually hit with stand-off strategic weapons and B2s first, that seems OK to me.
Sounds like it would work, and while it would take quite some time to develop, airframes last decades. OTOH, there are good practical reasons that 4 cm radar is used for targeting, and long-wave would be easy to fox. Still, it will eventually make the F-35s less effective against high-tech opponents (and I don't think humanity has quite grown out of that possibility).
So you think the only goal of a university is to teach "the prime concern"? How sad.
Regardless, there are many sorts of programming jobs, and at the big players there are still plenty of jobs doing systems and kernel-level tasks, not just code grinder jobs ("I'm writing an inventory database - in the cloud!"). As programmers start aging out of the workforce, with no/few universities teaching the low-level stuff any more, supply of qualified programmers will dwindle, while demand stays steady - seems like a worthwhile skill set to me!
You've repeatedly missed my point that you won't think to use bit-twiddling if you've never studied bit-twiddling. However easy it may be to learn, if you don't have the tool in your mental toolbox, you'll try to solve the problem with less-good tools.
Recursion is the same way - it's quite rare that it's useful it in business coding, but when it is your code will be vastly better for it. If, however, you're not comfortable with recursion, you'll instead solve the problem in some horribly-awkward way.
Schools teach plenty of recursion, as it's one of those nifty abstractions, but fall down on the grungy details that aren't so fun to teach (but are just as useful). Thus we in industry get stuck teaching the grungy details to fresh college hires, or worse you get entire codebases where no one had the right tools, and the result is a horrible mess.
Well, the enemy pilots might well understand from longer-wave radar (even current scanning radar) there are F-35s attacking, but need to get closer to get missile lock. Meanwhile, the F-35s locked at far greater distance, already have missiles on the way, and are leaving. But you can design a missile today that would just fly close enough to the F-35s to get a, then switch to terminal guidance (it's a common design strategy, in general) - but with existing radar and F-35s, you have to be content with the missile picking a nearby target at random, and no one's willing to do that today.
As far as Canada, hey, what can I say the "it's cheaper to just be nice to everyone" strategy seems to work for you guys, and for the moment the US has your backs. But do expect our military funding to keep falling to the point where you can't count on that, as the US eventually goes broke. Try not to look too appealing or loot-able when that happens.
If it's the case that the Russians and Chinese now have radar systems that remove that radar superiority, the F-35 now looks like even more of a gigantic waste of money
The F-35 was designed to be stealthy, not stealth. It doesn't need to be undetectable, as it's not a strategic bomber, it just needs to be able to get missile lock on it's foes before they get missile lock on the F-35. That doesn't seem like to change any time soon.
While any new military project whatsoever will be ridiculed as a colossal waste of money by the left ("it doesn't cost anything to just be nice to everyone!"), the main problem with the cost of most of the recent programs is a large R&D cost that isn't spread across enough planes/ships/whatever. I'm not the biggest fan of the F-35, but at least the idea of having one plane that will be used for many roles and by many allies keeps the per-unit cost from being insanely high - it's a wise procurement approach in a time of quickly falling defense budget.
And in none of those hits a bit fiddeling is involved on PHP o Java level? So none of the programmers there ever did an a = b | c & ^d.... Why should they?
All the big players have at least in part written their own OS variant, their own DB, their own device drivers, and so on. Large scale computing includes making the individual boxes work, and at the scale of a million servers it only makes sense to have full ownership of the entire stack.
I had lunch Tuesday with an Amazon coder who did bit bashing this week - in Java. Funny old world. From what I know about Amazon, you can't really make any generalizations about it, as different teams do things different ways far more than most big companies.
I used bit-level flags this year in C#, just because we had a great many of the objects in memory and didn't want to be wasteful with object size.
The default Java GC has been terrible for servers for most of its history - it wants to avoid GC until all memory is consumed, which is great for client code that exits before GC happens, and acceptable when you only have a couple of GB in any case. This causes 2 problems: GC can take too long to run, and you can't usefully monitor memory usage over time and alert when your app is nearly out of memory, as you get too many false positives.
G1 GC solves these problems nicely, but was only fully supported as of JDK 1.7 update 4.
This says it much better than I could. Most students carrying 100k in debt, most parents sending kids to school, most everyone really wants a job from college, not the "higher goal". Not that the "higher goal" is bad or wrong, but it's not what's being sold by universities. I think you'd get few takers at the price for what you see as the primary goal, but if everyone were honest I think it would all sort itself out, and most people would go to trade schools, and that would be best for everyone.
#define RED_FLAG 0x4 #define GREEN_FLAG 0x2 #define BLUE_FLAG 0x1 #define ANY_FLAG 0x7 // in some function
if (0 == foo & ANY_FLAG) return;
This looks like it testing whether any flag is set, so as to return early if there's no work to do. But it's not, instead it's kicking you right in the debugger.
I totally agree that your straw-man mischaracterization of my points was wrong. You have totally thrashed your opponent, and the ground is littered with straw.
The most clever solution to counting the 1 bits in a word (and once again fastest, though it wasn't for a decade or so) is best understood using octal.:) You must grok the "n%63" in fullness to see the light.
I think another part of the problem is that the schools are focused on making generic programmers for generic jobs to satisfy the demand from the corporate funders
That would be my dream utopia! Imagine, coming out of college with actual job offers. 30 years ago that was the norm for engineers from good schools, but is almost unheard of for kids today.
But the problems you describe come from the reverse of this: students are taught the academic abstractions, not the grungy details of how shit actually gets done. Were colleges more like trade schools, they'd be focusing specifically on stuff like debugging and memory allocation and optimization because that's useful to employers, instead of classes that stick to stuff that's easy to teach so that everyone gets a degree (which is U of Phoenix's goal).
Very few people understand the entire C++ standard well, though many people understand most of it. C++ probably has more bizarre dark corners than any other language.
Or you may just be wondering why if (0 == foo & MASK) doesn't do what you expected. I think that one burns everyone sooner or later.
Algorithms are great and all, but you won't get hired unless you can actually write code. The fresh college hires usually won't be designing complex systems until far enough into their career that they've forgotten college anyhow, but writing code happens immediately. A shocking % of graduates somehow didn't get that (if you know any non-scripting language, you're generally fine for most big employers, as they expect to teach you everything anyhow).
But TFS was specifically asking about the stuff Java schools don't teach. You can learn a lot using Java, but the holes are important.
Except that's just wrong, at least in Java. Java is sensitive to excessive object creation, and most developers eventually develop defensive habits against wasteful object creation, but you somehow have to figure out what the problem is and change how you code.
Also, in Java the garbage collector behavior matters. The default GC until recently was just terrible for long-running services, as it was optimized for client performance. How would you even know this was your problem, or choose a better GC, without some real understanding.
Now, sure, if you're a coder who only does scripting languages, we can only hope performance doesn't matter in the first place, but a otherwise complete lack of understanding of what's really going on in a computer will cripple your ability to diagnose/solve problems in production.
I think people take the view that Microsoft writes the operating system but fail to understand that Microsoft needs to hire someone to do this rather than rely on the shoe-repairing gnomes.
given the UI for windows 8 and server 2012, those shoe-repairing gnomes might be a better choice next time...
The shoe-repairing gnome theory has remarkable predictive power!
Slashdot Mirror
Well, the F22 was built for the same role as the F16 had initially: pure air-to-air combat, but now with supercruise to intercept at a distance. Nothing ever built even comes close. It's a damn shame we built so few of them, as the F35 can't hold a candle to it in that role, but with its operational issues it's understandable that it landed on the chopping block. (Seriously? Computers crashed when crossing the international date line? That's just embarrassing.)
We can't have multiple different aircraft when a cheaper alternative is one aircraft that does the jobs to almost the same level.
Well, that word "almost" is disputed by many. Plus it creates all the normal mono-culture risks: if it turns out to have some exploitable weakness, than all our planes will have that weakness. But given we won't be funding the military at "superpower" levels much longer (arguably we already don't), it will have to do.
Ultimately, the transition to combat drones is well underway, and I don't think it will end up being that big a deal if the F35 is second-rate. It's sort of the stopgap while we phase out piloted combat aircraft over the next 20 years or so.
"Horizon" is a fuzzy word. The APG-71 radar system, now deployed for 30 years, allows locking on to multiple targets at 230 miles as deployed (twice that as designed, but the antenna actually used isn't ideal). Pick what terminology you want, but I call that "long range".
The replacement for the F16 is the F22 (which sadly we have far to few of). The F16 was designed as an interceptor, a very different role than an air superiority fighter (and one that has become almost meaningless), but it works well enough. The F35 isn't in the same league as the F22 for clearing the skies, but since we're cutting every military budget everywhere, we can't have nice things.
The thing is, only the NFL makes money.
But by that analogy, there are any number of MOBAs, for example, so there are several choices of publishers. Sure, for a given exact rules set, there's only one, but again that's the same as pro sports.
Think of the NFL as "football's publisher" and you have the same situation - monopoly control over professional gameplay hasn't been a problem thus far. The difference, think, is that congresscritters care about the NFL, and won't let it go off the rails too far (and the government has poked its nose in from time to time). It will be interesting to see how that plays out with eSports - whether something like FIFA will come to be.
All that sounds right to me. However, F-35s are already visible enough to high-tech, high-power radar for situational awareness - they aren't B2-style stealth aircraft - if only marginally so. Going to longer-wave scanning radar seems to me to be the same as just throwing out 10x the radar energy to get the same result - still important, but not a difference in kind.
Given we usually hit with stand-off strategic weapons and B2s first, that seems OK to me.
Sounds like it would work, and while it would take quite some time to develop, airframes last decades. OTOH, there are good practical reasons that 4 cm radar is used for targeting, and long-wave would be easy to fox. Still, it will eventually make the F-35s less effective against high-tech opponents (and I don't think humanity has quite grown out of that possibility).
So you think the only goal of a university is to teach "the prime concern"? How sad.
Regardless, there are many sorts of programming jobs, and at the big players there are still plenty of jobs doing systems and kernel-level tasks, not just code grinder jobs ("I'm writing an inventory database - in the cloud!"). As programmers start aging out of the workforce, with no/few universities teaching the low-level stuff any more, supply of qualified programmers will dwindle, while demand stays steady - seems like a worthwhile skill set to me!
You've repeatedly missed my point that you won't think to use bit-twiddling if you've never studied bit-twiddling. However easy it may be to learn, if you don't have the tool in your mental toolbox, you'll try to solve the problem with less-good tools.
Recursion is the same way - it's quite rare that it's useful it in business coding, but when it is your code will be vastly better for it. If, however, you're not comfortable with recursion, you'll instead solve the problem in some horribly-awkward way.
Schools teach plenty of recursion, as it's one of those nifty abstractions, but fall down on the grungy details that aren't so fun to teach (but are just as useful). Thus we in industry get stuck teaching the grungy details to fresh college hires, or worse you get entire codebases where no one had the right tools, and the result is a horrible mess.
Well, the enemy pilots might well understand from longer-wave radar (even current scanning radar) there are F-35s attacking, but need to get closer to get missile lock. Meanwhile, the F-35s locked at far greater distance, already have missiles on the way, and are leaving. But you can design a missile today that would just fly close enough to the F-35s to get a, then switch to terminal guidance (it's a common design strategy, in general) - but with existing radar and F-35s, you have to be content with the missile picking a nearby target at random, and no one's willing to do that today.
As far as Canada, hey, what can I say the "it's cheaper to just be nice to everyone" strategy seems to work for you guys, and for the moment the US has your backs. But do expect our military funding to keep falling to the point where you can't count on that, as the US eventually goes broke. Try not to look too appealing or loot-able when that happens.
If it's the case that the Russians and Chinese now have radar systems that remove that radar superiority, the F-35 now looks like even more of a gigantic waste of money
The F-35 was designed to be stealthy, not stealth. It doesn't need to be undetectable, as it's not a strategic bomber, it just needs to be able to get missile lock on it's foes before they get missile lock on the F-35. That doesn't seem like to change any time soon.
While any new military project whatsoever will be ridiculed as a colossal waste of money by the left ("it doesn't cost anything to just be nice to everyone!"), the main problem with the cost of most of the recent programs is a large R&D cost that isn't spread across enough planes/ships/whatever. I'm not the biggest fan of the F-35, but at least the idea of having one plane that will be used for many roles and by many allies keeps the per-unit cost from being insanely high - it's a wise procurement approach in a time of quickly falling defense budget.
And in none of those hits a bit fiddeling is involved on PHP o Java level? So none of the programmers there ever did an a = b | c & ^d .... Why should they?
All the big players have at least in part written their own OS variant, their own DB, their own device drivers, and so on. Large scale computing includes making the individual boxes work, and at the scale of a million servers it only makes sense to have full ownership of the entire stack.
I had lunch Tuesday with an Amazon coder who did bit bashing this week - in Java. Funny old world. From what I know about Amazon, you can't really make any generalizations about it, as different teams do things different ways far more than most big companies.
I used bit-level flags this year in C#, just because we had a great many of the objects in memory and didn't want to be wasteful with object size.
The default Java GC has been terrible for servers for most of its history - it wants to avoid GC until all memory is consumed, which is great for client code that exits before GC happens, and acceptable when you only have a couple of GB in any case. This causes 2 problems: GC can take too long to run, and you can't usefully monitor memory usage over time and alert when your app is nearly out of memory, as you get too many false positives.
G1 GC solves these problems nicely, but was only fully supported as of JDK 1.7 update 4.
This says it much better than I could. Most students carrying 100k in debt, most parents sending kids to school, most everyone really wants a job from college, not the "higher goal". Not that the "higher goal" is bad or wrong, but it's not what's being sold by universities. I think you'd get few takers at the price for what you see as the primary goal, but if everyone were honest I think it would all sort itself out, and most people would go to trade schools, and that would be best for everyone.
#define RED_FLAG 0x4
// in some function
#define GREEN_FLAG 0x2
#define BLUE_FLAG 0x1
#define ANY_FLAG 0x7
if (0 == foo & ANY_FLAG) return;
This looks like it testing whether any flag is set, so as to return early if there's no work to do. But it's not, instead it's kicking you right in the debugger.
I totally agree that your straw-man mischaracterization of my points was wrong. You have totally thrashed your opponent, and the ground is littered with straw.
(Ewww! Octal!)
The most clever solution to counting the 1 bits in a word (and once again fastest, though it wasn't for a decade or so) is best understood using octal. :) You must grok the "n%63" in fullness to see the light.
I think another part of the problem is that the schools are focused on making generic programmers for generic jobs to satisfy the demand from the corporate funders
That would be my dream utopia! Imagine, coming out of college with actual job offers. 30 years ago that was the norm for engineers from good schools, but is almost unheard of for kids today.
But the problems you describe come from the reverse of this: students are taught the academic abstractions, not the grungy details of how shit actually gets done. Were colleges more like trade schools, they'd be focusing specifically on stuff like debugging and memory allocation and optimization because that's useful to employers, instead of classes that stick to stuff that's easy to teach so that everyone gets a degree (which is U of Phoenix's goal).
Very few people understand the entire C++ standard well, though many people understand most of it. C++ probably has more bizarre dark corners than any other language.
Or you may just be wondering why if (0 == foo & MASK) doesn't do what you expected. I think that one burns everyone sooner or later.
lgw: You don't need to know how they work in order to use them, but instead to know when to use them. E.g., reverse for hash functions.
a'o's: No, no, you don't need to know how they work in order to use them. Oh, I never thought about reverse for hash functions.
lgw: *headdesk*
Algorithms are great and all, but you won't get hired unless you can actually write code. The fresh college hires usually won't be designing complex systems until far enough into their career that they've forgotten college anyhow, but writing code happens immediately. A shocking % of graduates somehow didn't get that (if you know any non-scripting language, you're generally fine for most big employers, as they expect to teach you everything anyhow).
But TFS was specifically asking about the stuff Java schools don't teach. You can learn a lot using Java, but the holes are important.
Except that's just wrong, at least in Java. Java is sensitive to excessive object creation, and most developers eventually develop defensive habits against wasteful object creation, but you somehow have to figure out what the problem is and change how you code.
Also, in Java the garbage collector behavior matters. The default GC until recently was just terrible for long-running services, as it was optimized for client performance. How would you even know this was your problem, or choose a better GC, without some real understanding.
Now, sure, if you're a coder who only does scripting languages, we can only hope performance doesn't matter in the first place, but a otherwise complete lack of understanding of what's really going on in a computer will cripple your ability to diagnose/solve problems in production.
I think people take the view that Microsoft writes the operating system but fail to understand that Microsoft needs to hire someone to do this rather than rely on the shoe-repairing gnomes.
given the UI for windows 8 and server 2012, those shoe-repairing gnomes might be a better choice next time...
The shoe-repairing gnome theory has remarkable predictive power!
If you can get away with making it more negative, there's more to drain!