It is not CPU time, it is wall clock time. BTW, this is one reason why web applications currently suck, they are extremely slow on best and make you wait for seconds and sometimes even longer on worst.
I am certain that a few thousand lines of code like that, which is about average of C I have seen during last ~20 years, it would pass your audit.
There is exactly zero change that you would fix every single missing parenthesis. Besides, as pointed out, the missing parentheses has nothing to do with the problem.
Maybe the biggest problem with generics is that the "woods" gets hidden by too many "trees". Although I do not agree everything with http://www.artima.com/weblogs/viewpost.jsp?thread=299081 there is good example in the bottom. The beef, "map.sort.each", is hidden in the Java example by far too much litter.
As said by metamatic (above/sideways/how-should-I-describe-this) Java's first mistake was to have non-object types. The performance gain was minimal as optimizing compilers would have short-circuited most. E.g. Eiffel claimed 1-3% penalty at that time (when Java was born).
Then there is the Gödel's theorem. It states that no mathematical system can be "full" and "consistent". I believe it holds for programming languages too: either it is unusable or has some "inconsistent" semantics/behaviour.
Even more, design-by-committee problem: committee cannot make a clean design. Nowadays Java includes everything from kitchen sink to lambdas and soon parentheses and Prolog I fear. Sure it is nice to have function-almost-objects without class, but then so what?
You'll notice that I am no longer consistent, I hate extra verbosity that genericity incurs but then don't mind verbosity that full classes incur. Well, I do believe in "Gödel-extended-to-language":-)
Perhaps closures could have been put nicely into the Java when Java was designed first time. Now adding them into the soup is most likely going to make the language worse with almost no gain. Unless you consider as "gain" the fact that some language theorists gets a few brain-orgasms.
They should have left Java as it was and made a new language on top of JVM.
Obviously I would have put it quite differently. For example last example would "ensure" that getInfo returns a map which is "? extends sortedmap", keys are "? extends comparable", etc.
After those the beef (map.sort.each) would be a tiny percentage of the entire code. This does nut sum up, IMHO.
Code clarity certainly did not improve, and that is my main concern. The fact that generics cannot handle even "int" and work entirely on type erasure are "icing on the cake".
Maintenance is nightmare, just get hit by a code written someone who did not fully understand every item in Langer's FAQ and you are screwed. So it does not help at all that I learn it.
Trying to change an API you'll notice you have to rewrite entire application. A lot like "const" in C++, once you put it in somewhere, you need to put it everywhere until you notice you need to cast it away somewhere.
In theory generics help, but in practice they just make a mess of the language design.
Generics "killed" Java (well, was a huge mistake, though did not kill it).
Since then practically every new language has been "higher level". Designers noticed that generics solves trivial problems[1] with huge cost (code clarity, maintenance and education). Now they seem to be putting everything up lamdba calculus, logic programming and parentheses in. That leaves BNF to be integrated so that programmers can invent heir own favourite extension to the language.
[1] problems which are found in simplest unit tests and are easy to fix
I think it is good that government makes some rules how Facebook stores your personal information.
I have seen far too many cases of how not to do it, how not to store passwords, social security numbers, medical history, purchase history, etc.
Sure "feeding the starving children" is more important, but it does not mean I am willing to allow Google to do whatever it feels is best just because "I am not forced to deal with them".
How the offline test is going to tell you that only machines in the lab are going to be affected? How the test is going to tell you correct command order is "cp a a.bak; mv b a", not "mv; cp", or option should be "-i", not "-f"? Do you really make so extensive tests that you will notice all those corner cases? I very much doubt.
With GUI the computers in the lab can have different colour or visually separated. The colour difference can be dynamic, e.g. output of some database query[1]. With GUI you really cannot select "mv -f", only "cp -i" - at least not without physical feedback (colour, pop-up,...). With GUI you can see "do you want to overwrite 'b' with size X date Y with 'a' of size A date B", CLI gives "[Y/N]".
A good GUI is far superior to CLI.
[1] The point is, you can easily see that computers A, B & C are in the change group and "none outside this site" and X, Y & Z are not.
Sure you can accomplish most of those with CLI, but it is not that easy, intuitive and far more error-prone. And takes a helluva lot more time & testing.
I do believe in your numbers, I was just wondering why our numbers do not match.
And I found out most likely reason: you use imperial units, I use metric. For me U = W/(K*m^2), i.e. Watts per (kelvin * square meter). I do not know what units you use for R.
The size of windows really does not matter as much as you think.
You see, the size of the windows will be more than order of magnitude smaller than walls, floors and ceilings put together. So they cannot lose as much heat as everything else, unless you use crap ones or really huge ones.
Besides "low energy" house uses (depending on definition of "low energy") about half of the energy on water (shower, washing, etc).
In Finland "low energy" house has walls with U ~0.17, windows ~1.0 (I am not sure how to convert U to R). Anyway U < 0.2 is in practice far from easy, e.g. there may not be any leaks in the humidity barrier.
So you really can have big windows if you want. Sure they will be much more expensive than wall insulation, but $2000 each in a 200'000-300'000 house that hardly matters (prices in Finland for "typical" L-E window and house).
I'd say it is as humorous as recent Dilberts. That is, very boring.
Besides, his assumptions are utterly idiotic. For example windows are not that bad energy losers (U < 0.8 are available easily). If you have photovoltaics the colour of the roof hardly matters. Insulation costs next to nothing, unless you want to use more expensive ones (to keep it thin). Etc, see e.g. http://en.wikipedia.org/wiki/Low-energy_house
The Elo system does not depend on the premise "nothing affects the outcome of a game other than the skill of each player".
Sure it is modelled according to that, but in practise it is very untrue even for chess. There are a lot of examples where player A has won player B N out of M times although according to rating difference very different outcome should have happened.
The chess events are not similar, I have played a few and they do vary considerably (number of games per day, travel, lighting, temperature, players, mood,...).
Elo rating is much more meaningful than ratings used e.g. in snooker or tennis. No, I am not saying they should change their rating, it would not make much sense to change the culture of the game, after all the rating is just a bloody number.
Slashdot Mirror
It is not CPU time, it is wall clock time.
BTW, this is one reason why web applications currently suck, they are extremely slow on best and make you wait for seconds and sometimes even longer on worst.
I call bullshit.
I am certain that a few thousand lines of code like that, which is about average of C I have seen during last ~20 years, it would pass your audit.
There is exactly zero change that you would fix every single missing parenthesis.
Besides, as pointed out, the missing parentheses has nothing to do with the problem.
You do understand that the person who would win your citizenship would certainly sue you?
You must have seen this: http://www.theregister.co.uk/2010/11/29/wikileaks_shocker/
'The Yemeni government agreed to assume full responsibility for Sarah Palin in return for "substantial military aid".'
Partially, yes, mostly, no.
Maybe the biggest problem with generics is that the "woods" gets hidden by too many "trees". Although I do not agree everything with http://www.artima.com/weblogs/viewpost.jsp?thread=299081 there is good example in the bottom. The beef, "map.sort.each", is hidden in the Java example by far too much litter.
As said by metamatic (above/sideways/how-should-I-describe-this) Java's first mistake was to have non-object types. The performance gain was minimal as optimizing compilers would have short-circuited most. E.g. Eiffel claimed 1-3% penalty at that time (when Java was born).
Then there is the Gödel's theorem. It states that no mathematical system can be "full" and "consistent". I believe it holds for programming languages too: either it is unusable or has some "inconsistent" semantics/behaviour.
Even more, design-by-committee problem: committee cannot make a clean design. Nowadays Java includes everything from kitchen sink to lambdas and soon parentheses and Prolog I fear. Sure it is nice to have function-almost-objects without class, but then so what?
You'll notice that I am no longer consistent, I hate extra verbosity that genericity incurs but then don't mind verbosity that full classes incur. Well, I do believe in "Gödel-extended-to-language" :-)
Perhaps closures could have been put nicely into the Java when Java was designed first time. Now adding them into the soup is most likely going to make the language worse with almost no gain. Unless you consider as "gain" the fact that some language theorists gets a few brain-orgasms.
They should have left Java as it was and made a new language on top of JVM.
Actually (though I am not Ruby programmer), article http://www.artima.com/weblogs/viewpost.jsp?thread=299081 sums this up quite nicely.
Obviously I would have put it quite differently. For example last example would "ensure" that getInfo returns a map which is "? extends sortedmap", keys are "? extends comparable", etc.
After those the beef (map.sort.each) would be a tiny percentage of the entire code. This does nut sum up, IMHO.
Code clarity certainly did not improve, and that is my main concern.
The fact that generics cannot handle even "int" and work entirely on type erasure are "icing on the cake".
Maintenance is nightmare, just get hit by a code written someone who did not fully understand every item in Langer's FAQ and you are screwed. So it does not help at all that I learn it.
Trying to change an API you'll notice you have to rewrite entire application. A lot like "const" in C++, once you put it in somewhere, you need to put it everywhere until you notice you need to cast it away somewhere.
In theory generics help, but in practice they just make a mess of the language design.
Generics "killed" Java (well, was a huge mistake, though did not kill it).
Since then practically every new language has been "higher level". Designers noticed that generics solves trivial problems[1] with huge cost (code clarity, maintenance and education). Now they seem to be putting everything up lamdba calculus, logic programming and parentheses in. That leaves BNF to be integrated so that programmers can invent heir own favourite extension to the language.
[1] problems which are found in simplest unit tests and are easy to fix
Exactly. VNC is usable over low latency links, X is not.
I think it is good that government makes some rules how Facebook stores your personal information.
I have seen far too many cases of how not to do it, how not to store passwords, social security numbers, medical history, purchase history, etc.
Sure "feeding the starving children" is more important, but it does not mean I am willing to allow Google to do whatever it feels is best just because "I am not forced to deal with them".
How the offline test is going to tell you that only machines in the lab are going to be affected?
How the test is going to tell you correct command order is "cp a a.bak; mv b a", not "mv; cp", or option should be "-i", not "-f"? Do you really make so extensive tests that you will notice all those corner cases? I very much doubt.
With GUI the computers in the lab can have different colour or visually separated. The colour difference can be dynamic, e.g. output of some database query[1]. With GUI you really cannot select "mv -f", only "cp -i" - at least not without physical feedback (colour, pop-up, ...). With GUI you can see "do you want to overwrite 'b' with size X date Y with 'a' of size A date B", CLI gives "[Y/N]".
A good GUI is far superior to CLI.
[1] The point is, you can easily see that computers A, B & C are in the change group and "none outside this site" and X, Y & Z are not.
Sure you can accomplish most of those with CLI, but it is not that easy, intuitive and far more error-prone. And takes a helluva lot more time & testing.
The data of those who are not banned need not be collected. But alas, it was.
The (USA) media (and probably USA justice system too) will blame almost entirely on BP. ... yeah, you got it.
The UK media, OTOH, will
This is too easy.
I do believe in your numbers, I was just wondering why our numbers do not match.
And I found out most likely reason: you use imperial units, I use metric. For me U = W/(K*m^2), i.e. Watts per (kelvin * square meter). I do not know what units you use for R.
20 cm thick glass wool gives U=0.15. This is minimum as required by law (from this year on). Until now often less was used.
I am absolutely certain "basic walls" do not give U=0.08.
Typical value in Finland (2005) was 0.21 (for the wall).
You seem to be lost in the numbers.
The size of windows really does not matter as much as you think.
You see, the size of the windows will be more than order of magnitude smaller than walls, floors and ceilings put together. So they cannot lose as much heat as everything else, unless you use crap ones or really huge ones.
Besides "low energy" house uses (depending on definition of "low energy") about half of the energy on water (shower, washing, etc).
In Finland "low energy" house has walls with U ~0.17, windows ~1.0 (I am not sure how to convert U to R). Anyway U < 0.2 is in practice far from easy, e.g. there may not be any leaks in the humidity barrier.
So you really can have big windows if you want. Sure they will be much more expensive than wall insulation, but $2000 each in a 200'000-300'000 house that hardly matters (prices in Finland for "typical" L-E window and house).
Sure, best window is about same as worst wall. Which means you can have as big windows as you like (within a reason).
Insulation cost difference comparing "normal" and "low energy" house is minimal (~3%).
I'd say it is as humorous as recent Dilberts. That is, very boring.
Besides, his assumptions are utterly idiotic. For example windows are not that bad energy losers (U < 0.8 are available easily). If you have photovoltaics the colour of the roof hardly matters. Insulation costs next to nothing, unless you want to use more expensive ones (to keep it thin). Etc, see e.g. http://en.wikipedia.org/wiki/Low-energy_house
Or she changes her mind (and says so) after he does something she does not want. Which is rape.
Yes, it is legal. Goto http://www.illumos.org/.
For me the biggest reason for OpenSolaris is binary compatibility.
Whenever I update my Ubuntu I fear something will break. Quite often something (outside the distro) does break, especially when the kernel changes.
Sure those who are happy to use only those programs and hardware directly supported by Canonical may be happy, but I am not one of those.
The Elo system does not depend on the premise "nothing affects the outcome of a game other than the skill of each player".
Sure it is modelled according to that, but in practise it is very untrue even for chess. There are a lot of examples where player A has won player B N out of M times although according to rating difference very different outcome should have happened.
The chess events are not similar, I have played a few and they do vary considerably (number of games per day, travel, lighting, temperature, players, mood, ...).
Elo rating is much more meaningful than ratings used e.g. in snooker or tennis. No, I am not saying they should change their rating, it would not make much sense to change the culture of the game, after all the rating is just a bloody number.
So just because Taliban was even worse, Americans have the right to commit atrocities. Now I got it!
So now USA is going to relocate those who are in danger, right? Right?