www.digitalmars.com         C & C++   DMDScript  

digitalmars.D - GCJ vs. D

reply Jeremy <Jeremy_member pathlink.com> writes:
What do you think:

GCJ (native Java GNU compiler project) vs. DMD?

My thoughts:

* DMD is still much more faster and memory efficient, but GCJ v4 is closing the
gap
* GCJ allows you to compile Java as if it was just another language, but your
code can still be made into bytecode (so you ideally get the best of both
worlds)
* GCJ can interface with C/C++ (CNI)
* DMD lets you get away from everything-is-an-object which can be nice...

I think it is going to be harder for DMD to compete if such a strong native Java
compiler is making good progress...?
Mar 30 2006
next sibling parent reply David Medlock <noone nowhere.com> writes:
Jeremy wrote:
 What do you think:
 
 GCJ (native Java GNU compiler project) vs. DMD?
 
 My thoughts:
 
 * DMD is still much more faster and memory efficient, but GCJ v4 is closing the
 gap
 * GCJ allows you to compile Java as if it was just another language, but your
 code can still be made into bytecode (so you ideally get the best of both
 worlds)
 * GCJ can interface with C/C++ (CNI)
 * DMD lets you get away from everything-is-an-object which can be nice...
 
 I think it is going to be harder for DMD to compete if such a strong native
Java
 compiler is making good progress...?
 

If you were going to develop a 3D game, or a new scripting language, or some other low level software D beats Java easily. Not only that, D is superior as a language to Java, IMO. See: news://news.digitalmars.com:119/dvreco$hf1$1 digitaldaemon.com Java's templates are an absolute joke. For about 30% more typing we get homogenous containers and no casting in generic methods. Whoopee, Sun. They added aspects but shot down DBC. They still do not have any workable type inference at all. It took 10 years to add primitive boxing. The only place they win is support, users knowledgeable in the platform, and tools. -DavidM
Mar 30 2006
parent reply Kyle Furlong <kylefurlong gmail.com> writes:
David Medlock wrote:
 Jeremy wrote:
 What do you think:

 GCJ (native Java GNU compiler project) vs. DMD?

 My thoughts:

 * DMD is still much more faster and memory efficient, but GCJ v4 is 
 closing the
 gap
 * GCJ allows you to compile Java as if it was just another language, 
 but your
 code can still be made into bytecode (so you ideally get the best of both
 worlds)
 * GCJ can interface with C/C++ (CNI)
 * DMD lets you get away from everything-is-an-object which can be nice...

 I think it is going to be harder for DMD to compete if such a strong 
 native Java
 compiler is making good progress...?

If you were going to develop a 3D game, or a new scripting language, or some other low level software D beats Java easily. Not only that, D is superior as a language to Java, IMO. See: news://news.digitalmars.com:119/dvreco$hf1$1 digitaldaemon.com Java's templates are an absolute joke. For about 30% more typing we get homogenous containers and no casting in generic methods. Whoopee, Sun. They added aspects but shot down DBC. They still do not have any workable type inference at all. It took 10 years to add primitive boxing. The only place they win is support, users knowledgeable in the platform, and tools.

These three things are *NOT* trivial. They are actually essential to the success of the language.
 
 -DavidM

Mar 30 2006
parent David Medlock <noone nowhere.com> writes:
Kyle Furlong wrote:
 David Medlock wrote:
 
 Jeremy wrote:

 What do you think:

 GCJ (native Java GNU compiler project) vs. DMD?

 My thoughts:

 * DMD is still much more faster and memory efficient, but GCJ v4 is 
 closing the
 gap
 * GCJ allows you to compile Java as if it was just another language, 
 but your
 code can still be made into bytecode (so you ideally get the best of 
 both
 worlds)
 * GCJ can interface with C/C++ (CNI)
 * DMD lets you get away from everything-is-an-object which can be 
 nice...

 I think it is going to be harder for DMD to compete if such a strong 
 native Java
 compiler is making good progress...?

If you were going to develop a 3D game, or a new scripting language, or some other low level software D beats Java easily. Not only that, D is superior as a language to Java, IMO. See: news://news.digitalmars.com:119/dvreco$hf1$1 digitaldaemon.com Java's templates are an absolute joke. For about 30% more typing we get homogenous containers and no casting in generic methods. Whoopee, Sun. They added aspects but shot down DBC. They still do not have any workable type inference at all. It took 10 years to add primitive boxing. The only place they win is support, users knowledgeable in the platform, and tools.

These three things are *NOT* trivial. They are actually essential to the success of the language.
 -DavidM


If they were trivial I wouldnt have even mentioned them. Support requires money and an organization to take responsibility. Users knowledge is chicken-and-egg with the language popularity. Tools require, well, people to write them. -DavidM
Mar 31 2006
prev sibling parent reply Hong <Hong_member pathlink.com> writes:
I am doubtful that GCJ can close the gap much further

This page gives very good analysis on why Java is so damn slow, bytecode + VM is
not the only reason (which is the only one that GCJ removed).

http://www.jelovic.com/articles/why_java_is_slow.htm

Hong

In article <e0hvns$l0r$1 digitaldaemon.com>, Jeremy says...
What do you think:

GCJ (native Java GNU compiler project) vs. DMD?

My thoughts:

* DMD is still much more faster and memory efficient, but GCJ v4 is closing the
gap
* GCJ allows you to compile Java as if it was just another language, but your
code can still be made into bytecode (so you ideally get the best of both
worlds)
* GCJ can interface with C/C++ (CNI)
* DMD lets you get away from everything-is-an-object which can be nice...

I think it is going to be harder for DMD to compete if such a strong native Java
compiler is making good progress...?

Mar 30 2006
parent Tom <ihate spam.com> writes:
I'm a Java detractor in general (I do Java for some tasks where I work). 
  Though I really hate it very much because of it slowness (among other 
aspects), I'll have to say that the article is far from been made upon 
good scientific methods. Not so serious IMO.

Don't get me wrong, I totally sympathize with the article but if I'd 
have to embrace a serious position, I'd have to say it's too _light_ to 
be persuasive and satisfying.

--
Tom;

Hong escribió:
 I am doubtful that GCJ can close the gap much further
 
 This page gives very good analysis on why Java is so damn slow, bytecode + VM
is
 not the only reason (which is the only one that GCJ removed).
 
 http://www.jelovic.com/articles/why_java_is_slow.htm
 
 Hong
 
 In article <e0hvns$l0r$1 digitaldaemon.com>, Jeremy says...
 What do you think:

 GCJ (native Java GNU compiler project) vs. DMD?

 My thoughts:

 * DMD is still much more faster and memory efficient, but GCJ v4 is closing the
 gap
 * GCJ allows you to compile Java as if it was just another language, but your
 code can still be made into bytecode (so you ideally get the best of both
 worlds)
 * GCJ can interface with C/C++ (CNI)
 * DMD lets you get away from everything-is-an-object which can be nice...

 I think it is going to be harder for DMD to compete if such a strong native
Java
 compiler is making good progress...?


Mar 30 2006