Because you need a single atomic operation can both read from and write to the same memory location in one shot. Ah ha.... and a compiler can not generate that opcode ?
Hm... I must have missed something in the last years.
The question is, how much do you have to change the language to be a preferable choice for creating a kernel? Ha ha.
Why would you argue otherwise? Because your arguing is nonsense as well.
E.g. you did not mention it is a "log in script".
You said during log in, so I e.g. assume a dialog box on my PC asking for a username and password.
You are arguing about languages and claiming that people who use certain languages are by definition superiour to people using other languages.
I thought the same about VB developers... but I changed that mindset long ago before I started mainly to develop in Java.
P.S. Why did the log in script need a C++/Java program to do anything is beyond me anyway. So you want to argue you better had done the Java thing in C++ instead in "the right language" what ever your scripting language was?
E.g. my last company shipped Java based products to customers... those customers run Java 8. It does not help if I can upgrade to Java 9, but they can't.
And then again, there are probably API incompatibilities... e.g. running a Java 8 VM, but basically using deprecated Java 5 APIs, which are finally phased out in Java 9, hence you have to rework/fix the code base before you can do your magical "apt-get".
So each user was firing up a 70MB JVM when they logged in, which at 1000 users was 70GB of RAM. Which was quite a lot in 2004. Yes, and some other asshole did not realize you simply add the memory switches to the command line of the JVM to reduce it to 4MB or even 2MB, depending on what the complicated log in actually was doing. And then again, you could defer that to a single server, handling multiple log in attempts simultaneously and run everything in a single 70MB VM...
He wants to program multi threaded code on 64 CPU cores with 4x hyper threading in a network of minimum 16 hosts with POSIX only API in C. Obviously with sporadicly mixed in assembly. And all that because he is to stupid to type java -help into the console or does not grasp the then displayed command line options. I'm pretty sure he uses his own hand written index sequential DB which he pets since 1986, he tried once to migrate to berkley DB but failed miserable, and since then he feels miserable about any progress in Computer Science! Because it is not Science, it i engineering and only people with a proper engineering degree, as in electric engineering, should be allowed to even come close to a console. Because of the sparks... you know. Not to mix up with Apache Spark! Obviously also not to be mixed up with Spark Systems... oh, where was I?
Perhaps he belongs to the flolks who don't grasp what a classpath is? I once had a discussion with a C# freak, who claimed C# is so much better than Java because it has no $CLASSPATH... sigh...
They are known for their super fast Java VM, Zing, which supports better GCs and up and down scaling of memory and cores used: https://www.azul.com/products/...
Finally, their real business was "hardware VMs" with specialized multicore processors for native Java execution (well, I'm sure they do an on the fly byte code to machine code compilation, probably with dynamic optimization as in hotspot): https://www.azul.com/products/...
And yeah, spinlock implementations require assembly. No they don't. Why would they? Every language that compiles to machine code can be used. A no brainer...
You can hate the Java Eco System as much as you want: it is a language, and a vm and a set of libraries. If it lacks something you need for a specific reason, you always can upgrade the vm or the libraries or the compiler of the language.
There is no fundamental difference to C or C++ or assembly...
Your misinformation comes from a simple thing: you think for some reason Java/C# programmers are dummies. And because you don't use those languages you are superior, and hence the language you use is superior and hence you "imagine" stupid non existing limitations of the stuff you hate.
1. is very true and the main reason why C++ was adopted so slowly, especially on Windows. 2. I luckily never had such a bug, but I heard often about such issues.
That's like the guys in Germany who were getting paid to produce solar power. It was fine, but they noticed one company producing it at night. That was in Spain, not in Germany.
I've also heard of geniuses who go to WalMart (or wherever) buy thousands of dollars or merchandise and hand the clerk a million dollar bill. AND WANT THEIR CHANGE. If that happened to me, I would take the bill, tell him I have to consult my manager and check if it is genuine, and run to my car... Ah, well, the flaw is I usually come with my bicycle to work:(
(you don't, however, know if they have interacted with something or not). Of course you know. Technically a photon hitting a metal surface is not reflected. It is absorbed and recreated. Hence the entanglement with his "brother" breaks.
Other "ideas" about photons interacting with electrons, aka metallic surfaces, include that the photon changes its phase perpendicular to the surface. Aka it changes its polarization. Hence: it breaks entanglement with his "brother" or the "brother" changes polarization as well.
No idea about what you want to nitpick... both beams still only travel with the speed of light. And that is the only thing important. If you want: you can assume the detection is delayed by a time interval according to the distance the first photon has traveled... then you get rid of your "instant is impossible assumption".
Information is a relatively abstract thing. Of course it can be transported faster than light. Qunatum entanglement breaks instantly... if one photon gets forced to break its entanglement with the other one, you can see that at the other one. The trick is to have a way to keep an eye on a single photon... which is not that easy.
Another simple example is: you have a galactic mirror a few 100 million light years away, a pulsar or your torch light is moving a beam from left to right over it and you watch the reflection. That reflection can move with nearly any arbitrary speed...
An O(n) algorithm is an O(n) algorithm... always...
But keep believing that all O(n) perform the same when O() is a shitty, incomplete measurement of performance. I did not say that...
Both are O(n) yet the second runs AT LEAST 2x slower because the second has a memory access pattern that doesn't take advantage of how the L1/L2/L3 cache works. And? Two times slower is... O(n).
Perhaps you should reread what you think you know about O calculus. Even if it would be a factor of a million or "insert your number here" both algorithms would be a O(n). Welcome to the Real World. Theory matters... even if you don't grasp it.
And your evidence of plausibility is what, precisely? Perhaps you want to open a dictionary and check again what the words "plausible" and "evidence" mean.
If there was an civilization like my example: England around 1870/1900, somewhere e.g. in Indonesia, it is now 70m under the sea, hundreds of km off the coast, burried below sediments created over a time of 12000 years.
That is completely plausible. Is it true? No, I just made it up, obviously. It is called: "Gedankenexperiment"!
Slashdot Mirror
Ah, ha! That sounds funny!!
It is obvious that you can write that micro kernel in Java that emits native code ... so what is your point?
Because you need a single atomic operation can both read from and write to the same memory location in one shot. .... and a compiler can not generate that opcode ?
Ah ha
Hm ... I must have missed something in the last years.
The question is, how much do you have to change the language to be a preferable choice for creating a kernel?
Ha ha.
How many kernels are multithreaded aka reentrant?
Why would you argue otherwise?
Because your arguing is nonsense as well.
E.g. you did not mention it is a "log in script".
You said during log in, so I e.g. assume a dialog box on my PC asking for a username and password.
You are arguing about languages and claiming that people who use certain languages are by definition superiour to people using other languages.
I thought the same about VB developers ... but I changed that mindset long ago before I started mainly to develop in Java.
P.S.
Why did the log in script need a C++/Java program to do anything is beyond me anyway. So you want to argue you better had done the Java thing in C++ instead in "the right language" what ever your scripting language was?
My pub is my home
And the other pub is my living room!
You insensitive clod!
That is actually not the problem.
E.g. my last company shipped Java based products to customers ... those customers run Java 8. It does not help if I can upgrade to Java 9, but they can't.
And then again, there are probably API incompatibilities ... e.g. running a Java 8 VM, but basically using deprecated Java 5 APIs, which are finally phased out in Java 9, hence you have to rework/fix the code base before you can do your magical "apt-get".
So each user was firing up a 70MB JVM when they logged in, which at 1000 users was 70GB of RAM. Which was quite a lot in 2004. ...
Yes, and some other asshole did not realize you simply add the memory switches to the command line of the JVM to reduce it to 4MB or even 2MB, depending on what the complicated log in actually was doing.
And then again, you could defer that to a single server, handling multiple log in attempts simultaneously and run everything in a single 70MB VM
He wants to program multi threaded code on 64 CPU cores with 4x hyper threading in a network of minimum 16 hosts with POSIX only API in C. Obviously with sporadicly mixed in assembly. And all that because he is to stupid to type java -help into the console or does not grasp the then displayed command line options. I'm pretty sure he uses his own hand written index sequential DB which he pets since 1986, he tried once to migrate to berkley DB but failed miserable, and since then he feels miserable about any progress in Computer Science! Because it is not Science, it i engineering and only people with a proper engineering degree, as in electric engineering, should be allowed to even come close to a console. Because of the sparks ... you know. Not to mix up with Apache Spark! Obviously also not to be mixed up with Spark Systems ... oh, where was I?
Perhaps he belongs to the flolks who don't grasp what a classpath is? I once had a discussion with a C# freak, who claimed C# is so much better than Java because it has no $CLASSPATH ... sigh ...
No no no!!
Google for "Espresso Corretto", it does not need special chars :D
Indeed. Was about to write the same!
Azul is hardly a startup, it was founded 2002: https://en.wikipedia.org/wiki/...
They are known for their super fast Java VM, Zing, which supports better GCs and up and down scaling of memory and cores used: https://www.azul.com/products/...
Finally, their real business was "hardware VMs" with specialized multicore processors for native Java execution (well, I'm sure they do an on the fly byte code to machine code compilation, probably with dynamic optimization as in hotspot): https://www.azul.com/products/...
And yeah, spinlock implementations require assembly. ...
No they don't. Why would they? Every language that compiles to machine code can be used. A no brainer
You can hate the Java Eco System as much as you want: it is a language, and a vm and a set of libraries. If it lacks something you need for a specific reason, you always can upgrade the vm or the libraries or the compiler of the language.
There is no fundamental difference to C or C++ or assembly ...
Your misinformation comes from a simple thing: you think for some reason Java/C# programmers are dummies. And because you don't use those languages you are superior, and hence the language you use is superior and hence you "imagine" stupid non existing limitations of the stuff you hate.
1. is very true and the main reason why C++ was adopted so slowly, especially on Windows.
2. I luckily never had such a bug, but I heard often about such issues.
Yes, most likely it is, as this is an easy to use tool chain if you only have to write the compiler frontend.
However: llvm is written in c++ so rust has what ever feature the c++ developer wanted you to use.
This makes this statement in no way reasonable.
Rust links to C and C++ via LLVM, a statement like this makes more sense.
That's like the guys in Germany who were getting paid to produce solar power. It was fine, but they noticed one company producing it at night.
That was in Spain, not in Germany.
I've also heard of geniuses who go to WalMart (or wherever) buy thousands of dollars or merchandise and hand the clerk a million dollar bill. AND WANT THEIR CHANGE. ... :(
If that happened to me, I would take the bill, tell him I have to consult my manager and check if it is genuine, and run to my car
Ah, well, the flaw is I usually come with my bicycle to work
Google is your friend. The famoust one is even simply called JavaOS.
Must be hard to swallow that there are dozens of Java compilers that generate native code ... or why do you pretent to be ignorant? :P
You don't need to be elected ro be a minister.
However usually they are MPs, that is right.
Y2K is only an example if people still used 2 digit years in the 1990s.
Till the mid 1980s, no one really was aware to the problem.
(you don't, however, know if they have interacted with something or not).
Of course you know.
Technically a photon hitting a metal surface is not reflected.
It is absorbed and recreated. Hence the entanglement with his "brother" breaks.
Other "ideas" about photons interacting with electrons, aka metallic surfaces, include that the photon changes its phase perpendicular to the surface. Aka it changes its polarization. Hence: it breaks entanglement with his "brother" or the "brother" changes polarization as well.
No idea about what you want to nitpick ... both beams still only travel with the speed of light. And that is the only thing important. If you want: you can assume the detection is delayed by a time interval according to the distance the first photon has traveled ... then you get rid of your "instant is impossible assumption".
Information is a relatively abstract thing. ... if one photon gets forced to break its entanglement with the other one, you can see that at the other one. The trick is to have a way to keep an eye on a single photon ... which is not that easy.
Of course it can be transported faster than light. Qunatum entanglement breaks instantly
Another simple example is: you have a galactic mirror a few 100 million light years away, a pulsar or your torch light is moving a beam from left to right over it and you watch the reflection. That reflection can move with nearly any arbitrary speed ...
I know what I'm talking about.
And you don't know how to make an argument.
An O(n) algorithm is an O(n) algorithm ... always ...
But keep believing that all O(n) perform the same when O() is a shitty, incomplete measurement of performance. ...
I did not say that
Both are O(n) yet the second runs AT LEAST 2x slower because the second has a memory access pattern that doesn't take advantage of how the L1/L2/L3 cache works. ... O(n).
And? Two times slower is
Perhaps you should reread what you think you know about O calculus. Even if it would be a factor of a million or "insert your number here" both algorithms would be a O(n). Welcome to the Real World. Theory matters ... even if you don't grasp it.
It is irrelevant as you did not link different object files from different compilers (because in general you could not anyway).
You are right. It should have been modded +5 Interesting and not +2 Informative.
So: what is your mental problem?
And your evidence of plausibility is what, precisely?
Perhaps you want to open a dictionary and check again what the words "plausible" and "evidence" mean.
If there was an civilization like my example: England around 1870/1900, somewhere e.g. in Indonesia, it is now 70m under the sea, hundreds of km off the coast, burried below sediments created over a time of 12000 years.
That is completely plausible. Is it true? No, I just made it up, obviously. It is called: "Gedankenexperiment"!
Of course there is. ...
It is just not much