4 ms·

Java is a great platform for long-lasting robust apps. It is invaluable to both the business and the scientific community, providing stable code longevity and l

by api 16y ago

Java is a great platform for long-lasting robust apps. It is invaluable to both the business and the scientific community, providing stable code longevity and long term reusability beyond any other current platform. It provides near-C++ performance too if you tweak your code a bit. Yet here we are. Politics. Politics and business pissing contests. If they do ruin or derail Java, can you imagine the amount of lost capital? The amount of man-centuries of work that would at the very least now have to be ported? Of course, I don't think that will happen. The momentum is too large. But it could, and the fact that you have a bunch of salesman-type assholes threatening to destroy billions (or more) in value in order to measure the length of their dicks is sickening. Politics is sort of like farting. Yes, everyone does it. Yes, it is normal. But it smells, and the people who do it a lot are kind of gross.

gloob 16y ago
The amount of man-centuries of work that would at the very least now have to be ported? There are large numbers of Very Important Programs that are still written in some old dialect of Fortran. It's not like the Java 6 JVMs are going to disappear from the planet's face; businesses will just keep on using those.
- rbanffy 16y ago
  And the prophecy will then be fulfilled: Java will be the 21st century COBOL.
  - hsuresh 16y ago
    First COBOL of the 21st century. Am sure there will be lot more.
angstrom 16y ago
Was thinking this too, but I wouldn't forget Apache has also been working on Thrift: http://wiki.apache.org/thrift http://wiki.apache.org/thrift
rbanffy 16y ago
> It provides near-C++ performance too if you tweak your code a bit You mean, of course, tweaking the C++ side as well... ;-) I agree that in some cases you may get similar performance, but that usually means you are comparing superbly optimized Java code with horribly inefficient C++ code.
- dfox 16y ago
  Actually, I think that most people overestimate performance of C++. Using C++ the way authors intend almost invariably leads to gigantic binaries with large amounts of duplicated and also unnecessary autogenerated code. It is not too hard to believe that reasonable JIT can outperform that.
  - killedbydeath 16y ago
    C++ was designed in a way where only the features you use have overhead, so it is fairly easy to write the most computationally intense parts in C while using higher level abstractions everywhere else. This is something I missed a lot in Java when trying to write high performance code.
    - dfox 16y ago
      But most of this design was done 20 years ago and so used assumptions about CPU performance are mostly wrong for both modern CPUs and modern operating systems. For example overhead of calling virtual and non-virtual method across library boundary on ELF based system on reasonably modern CPU is essentially same.
      - beagle3 16y ago
        This is a very minor detail - so an optimization that was made 20 years ago is no longer relevant. Neither is it harmful or especially expensive.
  - Malus 16y ago
    It is not too difficult to generate small binaries from C++ code; modern C++ toolchains have link-time optimization (http://en.wikipedia.org/wiki/Link-time_optimization http://en.wikipedia.org/wiki/Link-time_optimization) and can optimize for size (-Os). Code size is rarely an issue in the C++ projects I work on (numerical and discrete event simulations).
pquerna 16y ago
I am trying to comprehend your comment. It seems like you are calling the Apache Software Foundation "bunch of salesman-type assholes threatening to destroy billions (or more) in value in order to measure the length of their dicks is sickening.". That kind of statement seems extremely delusional to the facts of how the ASF has interacted with the Java ecosystem for the last decade. The dispute is about a promise that Sun made -- and signed into with how the JCP was setup -- that a specification for a Java API, and the Testing of that API (TCK), are available open-enough to allow free and open source implementations. The ASF joined the JCP because it believed Sun would live up to this promise. Oracle supported the ASF position, before it bought Sun, because it wanted an more open Java language. But now that Oracle bought Sun, their position switched, and they are actively working to prevent other implementations of the Language from being created. This is not about politics and farting, this is about a contract and a promise that Sun made to the community.
- davidw 16y ago
  A more charitable interpretation of > It seems like you are calling the Apache Software Foundation "bunch of salesman-type assholes threatening to destroy billions (or more) in value in order to measure the length of their dicks is sickening.". might be that he's referring to Oracle. It's not very clear though.
  - Lewisham 16y ago
    That was my reading.
- api 16y ago
  I was referring to Oracle.
  - mark_l_watson 16y ago
    I thought you were :-) BTW, I am trying to build Harmony from source right now on OS X, just to see if it runs OK...

by api 16y ago

gloob 16y ago

The amount of man-centuries of work that would at the very least now have to be ported? There are large numbers of Very Important Programs that are still written in some old dialect of Fortran. It's not like the Java 6 JVMs are going to disappear from the planet's face; businesses will just keep on using those.

rbanffy 16y ago
And the prophecy will then be fulfilled: Java will be the 21st century COBOL.
- hsuresh 16y ago
  First COBOL of the 21st century. Am sure there will be lot more.

angstrom 16y ago

Was thinking this too, but I wouldn't forget Apache has also been working on Thrift: http://wiki.apache.org/thrift http://wiki.apache.org/thrift

rbanffy 16y ago

> It provides near-C++ performance too if you tweak your code a bit You mean, of course, tweaking the C++ side as well... ;-) I agree that in some cases you may get similar performance, but that usually means you are comparing superbly optimized Java code with horribly inefficient C++ code.

dfox 16y ago
Actually, I think that most people overestimate performance of C++. Using C++ the way authors intend almost invariably leads to gigantic binaries with large amounts of duplicated and also unnecessary autogenerated code. It is not too hard to believe that reasonable JIT can outperform that.
- killedbydeath 16y ago
  C++ was designed in a way where only the features you use have overhead, so it is fairly easy to write the most computationally intense parts in C while using higher level abstractions everywhere else. This is something I missed a lot in Java when trying to write high performance code.
  - dfox 16y ago
    But most of this design was done 20 years ago and so used assumptions about CPU performance are mostly wrong for both modern CPUs and modern operating systems. For example overhead of calling virtual and non-virtual method across library boundary on ELF based system on reasonably modern CPU is essentially same.
    - beagle3 16y ago
      This is a very minor detail - so an optimization that was made 20 years ago is no longer relevant. Neither is it harmful or especially expensive.
- Malus 16y ago
  It is not too difficult to generate small binaries from C++ code; modern C++ toolchains have link-time optimization (http://en.wikipedia.org/wiki/Link-time_optimization http://en.wikipedia.org/wiki/Link-time_optimization) and can optimize for size (-Os). Code size is rarely an issue in the C++ projects I work on (numerical and discrete event simulations).

pquerna 16y ago

I am trying to comprehend your comment. It seems like you are calling the Apache Software Foundation "bunch of salesman-type assholes threatening to destroy billions (or more) in value in order to measure the length of their dicks is sickening.". That kind of statement seems extremely delusional to the facts of how the ASF has interacted with the Java ecosystem for the last decade. The dispute is about a promise that Sun made -- and signed into with how the JCP was setup -- that a specification for a Java API, and the Testing of that API (TCK), are available open-enough to allow free and open source implementations. The ASF joined the JCP because it believed Sun would live up to this promise. Oracle supported the ASF position, before it bought Sun, because it wanted an more open Java language. But now that Oracle bought Sun, their position switched, and they are actively working to prevent other implementations of the Language from being created. This is not about politics and farting, this is about a contract and a promise that Sun made to the community.

davidw 16y ago
A more charitable interpretation of > It seems like you are calling the Apache Software Foundation "bunch of salesman-type assholes threatening to destroy billions (or more) in value in order to measure the length of their dicks is sickening.". might be that he's referring to Oracle. It's not very clear though.
- Lewisham 16y ago
  That was my reading.
api 16y ago
I was referring to Oracle.
- mark_l_watson 16y ago
  I thought you were :-) BTW, I am trying to build Harmony from source right now on OS X, just to see if it runs OK...