The fact that they haven't made much headway does not mean that they are not competing
That's exactly what it does mean. There is a big difference between competing and trying to.
With Java thats feasable, but pre-Java it was not so easy. There's a very big difference in the quality of c++ tools available for Windows and Unix.
Very true.
It can lead to using Windows servers to allow development with MS tools.
Only at the lowest-end scale, and technologies like DCOM weren't highly used as they were difficult to code for. DCOM (despite Microsoft's promises) never made it usably onto Unix, or any platform other than windows.
There is another issue here: For most serious projects the cuteness of the development tools is absolutely irrelevant to the choice of deployment platform or language. What matters is scalability, portability, protection of investment and the long term.
Java changes that, but guess what....NET pushes it back in the other direction.
No it doesn't. This is why: Java enables development using high-quality tools on any platform for deployment on any other..NET allows deployment using high-quality tools on Windows for deployment on Windows..NET doesn't even enforce the use of Windows Servers: it works perfectly well with Java server-side using SOAP and other protocols. There is a huge investment in, and deployment of, Java server-side..Net provides nothing that changes that, unless there is a political requirement in an organisation to only use Microsoft technology and deploy only on that platform.
Really? I haven't heard of any new projects being developed in COBOL.
I don't know about new projects, but look at job search sites. The COBOL job count is between 30%-40% of the C# job count, and many times the count for languages like Python. COBOL is very much alive, widely used, and lots of new code is written in it.
Again, I have to point out - we are not talking about 'currently'. The future is the issue.
Yes you were: you said Java could be made more popular, implying that it was not that popular now.
I believe It would be possible to use the Sun Java Desktop without using any Java.
Of course it would. But I don't get your point. Java Desktop is a Linux Desktop with Java and Java tools. That's kind of implied by the name.
The security tools aren't at all vital, but they are there to make things easier to manage for corporate systems.
Now by contrast how much of the Sun Java Desktop would work without the Linux kernel, the associated GNU tools and GPL'd software?
Well, it wouldn't. Java Desktop uses these tools. That is what the GPL allows. Would it be better if it ran only on a proprietary system, or is it better that Sun is helping to push Linux onto the desktop?
No prebuilt installer and most of your target audience can't deploy it.
Not true.
If they can copy a file, they can deploy it.
For web apps, you copy one single file (a WAR) into the directory of your application server on your website. That is all. No tweaking, no compiling, no configuration.
For other apps you download a jar file. On Windows: Double click jar file. On other platforms: java -jar jarfile No tweaking, no configuration.
There is nothing un-Open Source about this. The WAR file and jar files can contain the full source code if you wish.
You can deploy nice open-source java software so that users can compile it themselves. But the thing is... they don't have to compile it. Give them the class files packaged up right, and its a single command/click to run.
Not really. Microsoft's server section made its first loss last year, and their impact on the enterprise-scale server market (and I'm not including dedicated Exchange servers) has been minimal.
Development tools are factors in which platform to use. Superior development tools on Windows motivate people to choose to run windows.
Only for development purposes. A huge amount of Java development is done on Windows, but not for deployment on Windows.
COBOL was doing fine at one time too.
And still is. That is why it was one of the primary languages for.Net.
more attractive to developers.
You mean more attractive to Slashdotters and Open Sourcers. Java is currently about attractive to developers as its reasonably possible to be, if you look at the job market.
In the near future,.NET projects will outnumber Java projects simply on the momentum of legacy conversions alone. It makes Sun's Open Source timing critical. The clock is ticking, and they can't sit on their arses for too long before the OOP shift goes M$'s way and not theirs.
It doesn't matter. Its irrelevant to Java. It would matter if Java was being used in place of VB on the client side, but it isn't.
There is no such thing as 'The Software Market'. There are niches, which are quite separate.
Think of it like this. In transport there is the bicycle market and the truck market. There are huge numbers of bicycles produced, but that does not impact the truck market, or mean that the 'wheel' paradigm is 'going the bikes way'.
In the same way, there is the client (bicycle) side, and there is the server (truck) side. Until every aspect of.Net (not just the subset in Mono) becomes truly cross-platform, of proven quality, and opened up to vendor-independent design groups, its irrelevant to Java. Its not on the same machines, and not used for the same thing. Using.Net server side is just plain barmy: Why bother, when Java does even more in that situation, is free, is multi-vendor and portable?
I think really that it is the frameworks that will keep Java alive.
Good point, but there are several key features that will keep java not just alive but thriving.
1. Absolutely key: Java is multi-platform. Even Microsoft now seems to be admitting that they will have to share the server market. Java runs on all servers - even Microsoft. Using toolkits like SWT, Java can have the same access to the Windows API as any other Windows app. As far as I can see, the only reason to use.Net is political, not technical: you want to ensure your apps run only on Microsoft platforms. Java does what.Net does, but does the same everywhere.
2. Supported by lots of vendors. Even vendors who are in competition: Sun, IBM and HP, support Java and implement the standard.
It's not "run anywhere" if a large portion of the user base has good reasons not to install the JVM (for instance, its not worth the trouble of manually installing, and I don't like having more untrusted binaries on my system than I have to).
I understand your point, but Linux desktops are not a significant proportion of the user base (yet - I'm sure it will become so).
Java is the most widely used development language now. A recent survey showed that about half of Java developers use IDEs on Windows, and half use IDEs on Linux. That is a phenomenal amount of Java VM installations on Linux. You may have good reasons for not wanting to install Java, but they seem not to be shared by a substantial number of developers who work on Linux.
Python will evolve more in three years that Java in three years.
I would not try and predict anything about the IT business 3 years from now. I have no idea how much Python or Java will evolve, or how fast. This sound rather like wishful thinking. Java 1.5 is a dramatic step forward.
Is Java good enough in three years? Five years? 8 years?
I hope so, but how do you predict which language is good enough? Go with the most mainstream language with most vendor support.
Is the code I write today going to run on machines in 8 years? Will Java be around? That is the question.
Absolutely. No language before has had such industry-wide support. There have been dialects of other languages, like C++, but there is 1 Java standard (well, Ok, a few if you count things like J2ME) supported by all vendors (except MS).
If.Net is done right, and from the looks of it Microsoft has done it right, it will overwhelm Java for the simple reason that Microsoft has 95% of the desktop market.
No. Its about 80-85%. Otherwise, figures for Linux installations and MacOS/X + Windows would add up to more than 100%
Certainly future Microsoft OSes will run.Net out of the box. Actually.Net will probably run out-of-the-box on Linux AND Windows in the future, thanks to the work of what is now Novell. Java won't work out of the box except on a small handful of OSes (because the community rejects its license in favor of others.) If you can't see the demise of Java in the not-so-distant future, you must have your eyes shut.
You are confusing desktops with the entire IT market. Java works NOW on almost every OS. If its not installed, its a single click to download and install from java.net. Major computer VARs like Dell are bundling Java with desktops. However, for general use, this is irrelevant. Most java apps are accessed by web pages, and most java apps are deployed on servers: Desktop software is irrelevant to the Java market.
If you believe Java is in decline, you need to open your eyes and look at the real IT market, not the one you want to exist.
Yeah, just like linux is not directly competing with UNIX, and just like mySQL is no directly competing against Oracle.
But mysql is competing against Oracle because it can run on the same platforms..Net can't.
Without market share you cannot have market dominance. Without market dominance you cannot dictate your terms.
Java has the dominant market share for application development languages. Strangely, Sun aren't dictating terms - they gave Java to the JCP.
This is why Java will fail. It will cease to evolve quickly enough and will be passed by other, community-built languages.
Its doing fine so far. But, I guess you, in contrast to most of the IT industry, know better.
Java is built by a community. You are just opposing it because its not your community. If you want to join the Java community, you can. I know a fellow who has come up with some major feature designs which have been implemented in Java.
As far as serious mainframe development, I would be very leery of discounting python, especially since enormous strides are being made in making it run near-C speed.
Python is great, but Java has true power-level features such as security management and robust portable multithreading.
Of course, C is ultimately faster than Java, and might just be the ticket out.
if you want to go back to 70s-level technologies and coding. The Bus Error rules! Well, some of us want grown-up coding with decent features.
.Net isn't threatening Java. Microsoft say it is for marketing reasons..Net is replacing older development environments on Windows systems, such as VB6, VC++, ASP..Net is substantially client side. Until.Net is available for UNIX (a full featured system, not a catch-up clone like Mono) and for other serious server and mainframe systems, it won't compete directly with Java.
Yes - C# from Microsoft might pose a real threat to Java and make Sun redundant.
Only if there is a full-featured.Net available on AIX, Solaris, IBM Linux, HP Linux, mainframe systems (still a multi-billion dollar market), Mobile phones, most PDAs, embedded systems etc.
Until then, its a neat way to develop for Windows platforms, and to experiment with Mono on Linux.
Microsoft Visual C++ Borland C++ GNU C++ (different versions had incompatibilities) ARM C++ egcs++ VisualAge C++
Many is the time I have got out code from a few years ago and found it would not compile even with the same brand of compiler.
Won't bother me a bit when Java becomes just another language that GCC compiles to native code and it's bundled libraries are sitting in/usr/lib with the rest.
Well there goes the main benefit of Java: you ship compiled binary code and let your users and customers decide where it should be deployed. I guess either you don't believe in choice, or you want to support dozens of different platforms (including those which don't have '/usr/lib').
Well, i'm not sure how sun is making a ton of money off of java now.
They do it by selling software services.
Sun: "We'll sell you software stuff and services." Customer: "OK - what will it be written in?" Sun: "Java" Customer: "Cool - we already use java, its free, and we can use your software anywhere."
no connection to Java other than coming with a Java VM
And Sun Studio Java development tools. And Mozilla with full modern applet support. And Java security tools. And Java integration with Star Office to allow database connections and Office component development.
By the same reasoning Java Desktop has no connection to Linux other than including Linux...
Sun has no real value except for the ongoing license revenue from solaris hardware and software.
That's a very big 'except for'. Also, Sun now make a considerable amount of money from selling software services. Oh, and by the way, its 'Sparc' hardware. Solaris is an 'Operating System'.
Linux did its job and forced them into only high end servers
Funny, I thought Open Source was about the pleasure of writing quality code, comradeship between developers, and providing choice. I did not realise it was designed specifically to annoy Sun.
where there is - much less demand
There is?
- much stronger competition
But I thought you said that Linux was competition at the low end. How could Sun have been supposedly forced into 'only high end servers' without competition?
- commodity high end hardware
So?
- 'portable' dbms systems - Oracle 9 data is the same on aix, solaris, hp, etc
And has always been. Sun have always supported portability. That's why they went for UNIX decades ago and not a proprietary closed system like IBM and HP used to have. Sun got there first. When they started with UNIX, they published open standards for everyone to use, such as NFS. They allowed other manufacturers to use their Sparc designs. Sun realised that competition and portability are good: it means that competitors software can run on your systems.
Open sourcing java wouldn't really hurt them, and god knows java could use it.
Why?
Is Java suffering at all due to lack of demand? No.
Is the Java licencing restricting its implementation on different platforms? No.
Is Java on Linux suffering as a result of this licencing? No - Linux is one of the main deployment platforms for Java.
Is the demand for Java in the job market decreasing? No.
I fully support open sourcing Java, but it does not take much understanding of the IT industry to realise that Java certainly doesn't need open sourcing - its phenomenally successful.
The test was meant to show how expensive object creation is.
The original post stated that Java was generally slow, and said nothing about object creation.
An algorithm using object creation in C++ is a lot slower than an algorithm which uses raw memory in C. But that was not the point.
I've said that MONO executes java more efficiently than Sun's JAVA, nothing else.
Yes you have.
Its basic math.
Sun's Java executes typical code at within, say 20% of C/C++ speed. Note the word 'typical'. This means code in general.
So, Java time to run (J) = C++ time to run (C) * 1.2
if Mono (M) is much (say double) the speed of java... then we have the following equations:
J = C * 1.2 M = J / 2
To compare Mono and C...
M = C * 1.2 / 2
So
M = C * 0.6
Your statement implies that Mono is nearly twice as fast as native C++.
You may say this is nonsense. I agree with you! The point is that you can always select a particular benchmark to support a particular point of view. If you pick expression matching, for example, you can often find that Perl is much faster than C++. Does this mean Perl is generally faster than C++? Of course not! In the same way, Mono does not typically run Java significantly faster than Sun's JVM.
When the function was called several million times (so the program ran long enough for some good optimisation to kick in) The Sun JVM gave performance within 10%-20% of optimised GCC. the IBM VM was 10% faster than GCC.
THE POINT IS THAT MONO EXECUTES THE JAVA CODE MUCH(!) FASTER THAN THE SUN VM IMPLEMENTATION!
Impossible. The Sun VM implementation routinely gives speeds of within 10%-20% of compiled C or C++ for equivalent Java code. You would have to be also saying that Mono is 'MUCH(!)' faster than C or C++.
I too am fixing to being developing a new desktop gui app. I first thought 'Java' because I like cross platformness, however, working with Swing is a bitch.
Well don't use Swing then. There are plenty of other options: SWT, BISS-AWT.
Well, the benchmark tests the VM implementation, not the quality of the JIT optimization.
The quality of the JIT is a major part of the VM implementation.
Here's the C implementation of the parent's java code.
That is not equivalent. There is no memory allocation (equivalent to the creation of all those Integers).
I've run it on my machine and it is about 40 times faster than the java code.
Well, of course it does - there is no object creation.
Here is Java equivalent to your C:
-------------------- package net.parkplatz.web;
import java.io.*;
public class Test {
public final static int size = 1000000;
public static int [] vec = new int [size];
static int I;
public static int func() {
int i, count;
for(count=0; countSummary: Java is not much faster or slower than C++ code, but the implementations vary.
That is exactly what I said. The original post said Java was 26% of the speed of C++. This is nonsense.
Slashdot Mirror
The fact that they haven't made much headway does not mean that they are not competing
.NET allows deployment using high-quality tools on Windows for deployment on Windows. .NET doesn't even enforce the use of Windows Servers: it works perfectly well with Java server-side using SOAP and other protocols. There is a huge investment in, and deployment of, Java server-side. .Net provides nothing that changes that, unless there is a political requirement in an organisation to only use Microsoft technology and deploy only on that platform.
That's exactly what it does mean. There is a big difference between competing and trying to.
With Java thats feasable, but pre-Java it was not so easy. There's a very big difference in the quality of c++ tools available for Windows and Unix.
Very true.
It can lead to using Windows servers to allow development with MS tools.
Only at the lowest-end scale, and technologies like DCOM weren't highly used as they were difficult to code for. DCOM (despite Microsoft's promises) never made it usably onto Unix, or any platform other than windows.
There is another issue here: For most serious projects the cuteness of the development tools is absolutely irrelevant to the choice of deployment platform or language. What matters is scalability, portability, protection of investment and the long term.
Java changes that, but guess what....NET pushes it back in the other direction.
No it doesn't. This is why: Java enables development using high-quality tools on any platform for deployment on any other.
Really? I haven't heard of any new projects being developed in COBOL.
I don't know about new projects, but look at job search sites. The COBOL job count is between 30%-40% of the C# job count, and many times the count for languages like Python. COBOL is very much alive, widely used, and lots of new code is written in it.
Again, I have to point out - we are not talking about 'currently'. The future is the issue.
Yes you were: you said Java could be made more popular, implying that it was not that popular now.
I believe It would be possible to use the Sun Java Desktop without using any Java.
Of course it would. But I don't get your point. Java Desktop is a Linux Desktop with Java and Java tools. That's kind of implied by the name.
The security tools aren't at all vital, but they are there to make things easier to manage for corporate systems.
Now by contrast how much of the Sun Java Desktop would work without the Linux kernel, the associated GNU tools and GPL'd software?
Well, it wouldn't. Java Desktop uses these tools. That is what the GPL allows. Would it be better if it ran only on a proprietary system, or is it better that Sun is helping to push Linux onto the desktop?
And your Java 1.0 code builds perfectly on 1.4?
Yes. Java works like that.
No prebuilt installer and most of your target audience can't deploy it.
Not true.
If they can copy a file, they can deploy it.
For web apps, you copy one single file (a WAR) into the directory of your application server on your website. That is all. No tweaking, no compiling, no configuration.
For other apps you download a jar file.
On Windows: Double click jar file.
On other platforms: java -jar jarfile
No tweaking, no configuration.
There is nothing un-Open Source about this. The WAR file and jar files can contain the full source code if you wish.
You can deploy nice open-source java software so that users can compile it themselves. But the thing is... they don't have to compile it. Give them the class files packaged up right, and its a single command/click to run.
The platforms themselves are competing.
.Net.
Not really. Microsoft's server section made its first loss last year, and their impact on the enterprise-scale server market (and I'm not including dedicated Exchange servers) has been minimal.
Development tools are factors in which platform to use. Superior development tools on Windows motivate people to choose to run windows.
Only for development purposes. A huge amount of Java development is done on Windows, but not for deployment on Windows.
COBOL was doing fine at one time too.
And still is. That is why it was one of the primary languages for
more attractive to developers.
You mean more attractive to Slashdotters and Open Sourcers. Java is currently about attractive to developers as its reasonably possible to be, if you look at the job market.
Exactly.
But you did. You stated that Python would evolve faster than Java. Surely that is a prediction?
In the near future, .NET projects will outnumber Java projects simply on the momentum of legacy conversions alone. It makes Sun's Open Source timing critical. The clock is ticking, and they can't sit on their arses for too long before the OOP shift goes M$'s way and not theirs.
.Net (not just the subset in Mono) becomes truly cross-platform, of proven quality, and opened up to vendor-independent design groups, its irrelevant to Java. Its not on the same machines, and not used for the same thing. Using .Net server side is just plain barmy: Why bother, when Java does even more in that situation, is free, is multi-vendor and portable?
It doesn't matter. Its irrelevant to Java. It would matter if Java was being used in place of VB on the client side, but it isn't.
There is no such thing as 'The Software Market'. There are niches, which are quite separate.
Think of it like this. In transport there is the bicycle market and the truck market. There are huge numbers of bicycles produced, but that does not impact the truck market, or mean that the 'wheel' paradigm is 'going the bikes way'.
In the same way, there is the client (bicycle) side, and there is the server (truck) side. Until every aspect of
I think really that it is the frameworks that will keep Java alive.
.Net is political, not technical: you want to ensure your apps run only on Microsoft platforms. Java does what .Net does, but does the same everywhere.
Good point, but there are several key features that will keep java not just alive but thriving.
1. Absolutely key: Java is multi-platform. Even Microsoft now seems to be admitting that they will have to share the server market. Java runs on all servers - even Microsoft. Using toolkits like SWT, Java can have the same access to the Windows API as any other Windows app. As far as I can see, the only reason to use
2. Supported by lots of vendors. Even vendors who are in competition: Sun, IBM and HP, support Java and implement the standard.
It's not "run anywhere" if a large portion of the user base has good reasons not to install the JVM (for instance, its not worth the trouble of manually installing, and I don't like having more untrusted binaries on my system than I have to).
I understand your point, but Linux desktops are not a significant proportion of the user base (yet - I'm sure it will become so).
Java is the most widely used development language now. A recent survey showed that about half of Java developers use IDEs on Windows, and half use IDEs on Linux. That is a phenomenal amount of Java VM installations on Linux. You may have good reasons for not wanting to install Java, but they seem not to be shared by a substantial number of developers who work on Linux.
Python will evolve more in three years that Java in three years.
I would not try and predict anything about the IT business 3 years from now. I have no idea how much Python or Java will evolve, or how fast. This sound rather like wishful thinking. Java 1.5 is a dramatic step forward.
Is Java good enough in three years? Five years? 8 years?
I hope so, but how do you predict which language is good enough? Go with the most mainstream language with most vendor support.
Is the code I write today going to run on machines in 8 years? Will Java be around? That is the question.
Absolutely. No language before has had such industry-wide support. There have been dialects of other languages, like C++, but there is 1 Java standard (well, Ok, a few if you count things like J2ME) supported by all vendors (except MS).
If .Net is done right, and from the looks of it Microsoft has done it right, it will overwhelm Java for the simple reason that Microsoft has 95% of the desktop market.
.Net out of the box. Actually .Net will probably run out-of-the-box on Linux AND Windows in the future, thanks to the work of what is now Novell. Java won't work out of the box except on a small handful of OSes (because the community rejects its license in favor of others.) If you can't see the demise of Java in the not-so-distant future, you must have your eyes shut.
No. Its about 80-85%. Otherwise, figures for Linux installations and MacOS/X + Windows would add up to more than 100%
Certainly future Microsoft OSes will run
You are confusing desktops with the entire IT market. Java works NOW on almost every OS. If its not installed, its a single click to download and install from java.net. Major computer VARs like Dell are bundling Java with desktops. However, for general use, this is irrelevant. Most java apps are accessed by web pages, and most java apps are deployed on servers: Desktop software is irrelevant to the Java market.
If you believe Java is in decline, you need to open your eyes and look at the real IT market, not the one you want to exist.
Yeah, just like linux is not directly competing with UNIX, and just like mySQL is no directly competing against Oracle.
.Net can't.
But mysql is competing against Oracle because it can run on the same platforms.
Without market share you cannot have market dominance. Without market dominance you cannot dictate your terms.
Java has the dominant market share for application development languages. Strangely, Sun aren't dictating terms - they gave Java to the JCP.
This is why Java will fail. It will cease to evolve quickly enough and will be passed by other, community-built languages.
Its doing fine so far. But, I guess you, in contrast to most of the IT industry, know better.
Java is built by a community. You are just opposing it because its not your community. If you want to join the Java community, you can. I know a fellow who has come up with some major feature designs which have been implemented in Java.
As far as serious mainframe development, I would be very leery of discounting python, especially since enormous strides are being made in making it run near-C speed.
Python is great, but Java has true power-level features such as security management and robust portable multithreading.
Of course, C is ultimately faster than Java, and might just be the ticket out.
if you want to go back to 70s-level technologies and coding. The Bus Error rules! Well, some of us want grown-up coding with decent features.
.Net isn't threatening Java. Microsoft say it is for marketing reasons. .Net is replacing older development environments on Windows systems, such as VB6, VC++, ASP. .Net is substantially client side. Until .Net is available for UNIX (a full featured system, not a catch-up clone like Mono) and for other serious server and mainframe systems, it won't compete directly with Java.
Yes - C# from Microsoft might pose a real threat to Java and make Sun redundant.
.Net available on AIX, Solaris, IBM Linux, HP Linux, mainframe systems (still a multi-billion dollar market), Mobile phones, most PDAs, embedded systems etc.
Only if there is a full-featured
Until then, its a neat way to develop for Windows platforms, and to experiment with Mono on Linux.
Oh bull. How many incompatible forks of C++
/usr/lib with the rest.
Are you serious?
Microsoft Visual C++
Borland C++
GNU C++ (different versions had incompatibilities)
ARM C++
egcs++
VisualAge C++
Many is the time I have got out code from a few years ago and found it would not compile even with the same brand of compiler.
Won't bother me a bit when Java becomes just another language that GCC compiles to native code and it's bundled libraries are sitting in
Well there goes the main benefit of Java: you ship compiled binary code and let your users and customers decide where it should be deployed. I guess either you don't believe in choice, or you want to support dozens of different platforms (including those which don't have '/usr/lib').
thus fulfilling its original purpose: write once, run anywhere.
It already does that. The matter of open source is irrelevant.
Well, i'm not sure how sun is making a ton of money off of java now.
They do it by selling software services.
Sun: "We'll sell you software stuff and services."
Customer: "OK - what will it be written in?"
Sun: "Java"
Customer: "Cool - we already use java, its free, and we can use your software anywhere."
no connection to Java other than coming with a Java VM
And Sun Studio Java development tools.
And Mozilla with full modern applet support.
And Java security tools.
And Java integration with Star Office to allow database connections and Office component development.
By the same reasoning Java Desktop has no connection to Linux other than including Linux...
Sun has no real value except for the ongoing license revenue from solaris hardware and software.
That's a very big 'except for'. Also, Sun now make a considerable amount of money from selling software services. Oh, and by the way, its 'Sparc' hardware. Solaris is an 'Operating System'.
Linux did its job and forced them into only high end servers
Funny, I thought Open Source was about the pleasure of writing quality code, comradeship between developers, and providing choice. I did not realise it was designed specifically to annoy Sun.
where there is
- much less demand
There is?
- much stronger competition
But I thought you said that Linux was competition at the low end. How could Sun have been supposedly forced into 'only high end servers' without competition?
- commodity high end hardware
So?
- 'portable' dbms systems - Oracle 9 data is the same on aix, solaris, hp, etc
And has always been. Sun have always supported portability. That's why they went for UNIX decades ago and not a proprietary closed system like IBM and HP used to have. Sun got there first. When they started with UNIX, they published open standards for everyone to use, such as NFS. They allowed other manufacturers to use their Sparc designs. Sun realised that competition and portability are good: it means that competitors software can run on your systems.
Open sourcing java wouldn't really hurt them, and god knows java could use it.
Why?
Is Java suffering at all due to lack of demand?
No.
Is the Java licencing restricting its implementation on different platforms?
No.
Is Java on Linux suffering as a result of this licencing?
No - Linux is one of the main deployment platforms for Java.
Is the demand for Java in the job market decreasing?
No.
I fully support open sourcing Java, but it does not take much understanding of the IT industry to realise that Java certainly doesn't need open sourcing - its phenomenally successful.
The test was meant to show how expensive object creation is.
The original post stated that Java was generally slow, and said nothing about object creation.
An algorithm using object creation in C++ is a lot slower than an algorithm which uses raw memory in C. But that was not the point.
I've said that MONO executes java more efficiently than Sun's JAVA, nothing else.
Yes you have.
Its basic math.
Sun's Java executes typical code at within, say 20% of C/C++ speed. Note the word 'typical'. This means code in general.
So, Java time to run (J) = C++ time to run (C) * 1.2
if Mono (M) is much (say double) the speed of java... then we have the following equations:
J = C * 1.2
M = J / 2
To compare Mono and C...
M = C * 1.2 / 2
So
M = C * 0.6
Your statement implies that Mono is nearly twice as fast as native C++.
You may say this is nonsense. I agree with you! The point is that you can always select a particular benchmark to support a particular point of view. If you pick expression matching, for example, you can often find that Perl is much faster than C++. Does this mean Perl is generally faster than C++? Of course not! In the same way, Mono does not typically run Java significantly faster than Sun's JVM.
Google for "The Great Language Shootout"
I did. Its out of date (frozen in 2001) and the website says:
"It was never finished and contains errors. It should not be relied on as a true inter-language performance comparison."
The method/function to do the calulation is shown.
Java:C:When the function was called several million times (so the program ran long enough for some good optimisation to kick in) The Sun JVM gave performance within 10%-20% of optimised GCC. the IBM VM was 10% faster than GCC.
THE POINT IS THAT MONO EXECUTES THE JAVA CODE MUCH(!) FASTER THAN THE SUN VM IMPLEMENTATION!
Impossible. The Sun VM implementation routinely gives speeds of within 10%-20% of compiled C or C++ for equivalent Java code. You would have to be also saying that Mono is 'MUCH(!)' faster than C or C++.
I too am fixing to being developing a new desktop gui app. I first thought 'Java' because I like cross platformness, however, working with Swing is a bitch.
Well don't use Swing then. There are plenty of other options: SWT, BISS-AWT.
Well, the benchmark tests the VM implementation, not the quality of the JIT optimization.
The quality of the JIT is a major part of the VM implementation.
Here's the C implementation of the parent's java code.
That is not equivalent. There is no memory allocation (equivalent to the creation of all those Integers).
I've run it on my machine and it is about 40 times faster than the java code.
Well, of course it does - there is no object creation.
Here is Java equivalent to your C:
--------------------
package net.parkplatz.web;
import java.io.*;
public class Test {
public final static int size = 1000000;
public static int [] vec = new int [size];
static int I;
public static int func() {
int i, count;
for(count=0; countSummary: Java is not much faster or slower than C++ code, but the implementations vary.
That is exactly what I said. The original post said Java was 26% of the speed of C++. This is nonsense.