I was looking at this earlier, and it seemed pretty good. The only problem with Java is the much higher-level language: slower. Are there any stats comparing Java to C to ASM?

The powerful thing about java isn't it's speed. It's its portability. You could take a java program from a PC to a Macintosh with no problems at all, and you wouldn't even have to recompile the language because of java's "half-compling" method and byte code. I know of no other programming language that could do that.

Any language that can create bytecode for an interpretor. As Java programs still needs a Java interpretor, they are no more portable than a Perl, Ruby or Python program. A C program written in ANSI C is also portable, all you need is to compile it for the target platform.
Don't buy into the Java hype. It's nothing new. Just corporate (Sun) packaging of an old concept. Java programs are not more portable than any other language, given you enforce portability guidelines. It is possible to write platform-specific, not-portable, Java as well.

and the least portable is visual basic .net becasue you need windows 98 SE or up and you need a interpriter

Actually, Microsoft is working on a .NET framework for both Mac OS X and Linux (kernel 2.4+).

Not only Microsoft...
http://www.go-mono.com

Also, http://dotgnu.org/

Ah, good point. I'd completely forgotten about those.

For Linux? Really? Do you have a pointer to this information?

Last I checked there was just Rotor (which is open source, IIRC), but that one doesn't run on linux (BSD only) and is not complete.

Apparently I was wrong... the ones Kevin and Chivo pointed out (http://www.go-mono.com, http://dotgnu.org/) were what I was remembering. Sorry about the confusion.

Yeah, I can really see Micro$oft developing much of anything for the Linux platform. Besides, would any self-respecting Linux hacker actually use it?

Yeah, it's just for BSD, not for Linux. However, I read an article recently by a guy who tried to get Rotor to run on Linux through BSD emulation (more or less). I didn't read it all the way through, so I don't know if he succeeded.

(From what I remember, that guy hang-glides and pointed out that a "rotor" during hang-gliding is something that leads to a crash. :-)

Why is Microsoft doing anything for Linux or Mac

I don't know about linux, but Microsoft has been developing software for Apple-owned platforms for longer than it has been developing for Intel-based (80x86/8088) platforms.
Multiplan was designed for the Apple II; Mac Office is being developed constantly (or so they tell us)

The way I think of it is, Java isn't portable. You need JVM. I could say that C++, or any other language, for that matter, is portable. All you need are the correct emulators/interpreters/etc.

er.. no. With Java you need a JVM to even run the program, with C, C++, and other compiled languages, you only need to compile it once, and just use that binary on another computer of the same type(architecture). ...and OS-type, something compiled under linux may have a hard time running on Windows...

Then again, sometimes things compiled under a Linux kernel run better than things compiled under windows (assuming the right libraries are used) on a windows system.

Right. Java is actually a relatively slow language since its code has to be read by the JVM interpreter. That's one reason why I don't like it as much.

Anyway, have I seen this program before? I remember a loooooooong time ago, there was some Java->ASM project out there. Is this the same one?

You'd be surprised how fast java is because of the just in time (jit) compiler. The only problem with creating java applications is that the user interface is slow and they require a lot of memory for jit.

When you'll take a look at the link you'll see that java isn't as slow as most people think. Especially when certains sections of code are executed a lot.

This is a very bad benchmark. See:
http://gcc.gnu.org/ml/gcc/ ...
... 2004-02/msg00704.html
... 2004-02/msg00706.html
... 2004-02/msg00716.html

Moka doesn't even use a VM, AFAIK. It looks like Moka translates the Java program to a C program and compiles it with TIGCC.

While Moka programs wouldn't be BLAZING fast (due to abstraction layers etc.), they wouldn't be CRAWLING either. It's probably still tens or hundreds of times faster than TI-BASIC, though.

Someone ported Waba to the Ti-89/92+ a while back.
Waba is a JVM with a bunch of libraries designed for embedded devices, and runs on PalmOS and WinCE (PocketPC).
However, as a classic JVM it looked terribly slow, and a mouse interface isn't exactly suited for the calcs...
ticalc news article dated 6 June 2001.

Waba for 68K calcs was released while I was working on Moka. I was quite impressed (i never tough that a real bytecode interpreter could ever run on a TI-89) and I recognize that Waba is a great piece of software. But I was deceived by two reality : first, the speed was nearer to BASIC than to C (it was inevitable because waba program are compiled to bytecode, not machine language) and the release I tried featured only a very limited set of classes (no file or link I/O support, only the graphic classes needed to run the Scribble application were implemented and so on). Moka, although not as ‘perfect’ as Waba as a port of the Java language, is more suited for calculator because Moka program are converted in C to be compiled (a control structure such as a for is as fast in Moka as in C) and Moka offers at once an extended API and a complete interoperability with C.