On Mon, 18 Mar 2002, der Mouse wrote:

> > I don't think you understand what I'm saying.  When you get a Java
> > program, it's in bits and bytes, including all the fp numbers already
> > in ieee format.
> 
> Only if you're stuck with bytecode rather than source.

Which is what I said, and how much Java code is available.

> But even then, you don't need an "ieee floating point package", just
> converstion from IEEE to VAX.  (And back, if you want to generate
> bytecode compatible with IEEE JVMs.)

Which also is what I said. And storing in ieee is neccesary, or the next
read will screw you up as well.

> > Or are you saying that you will port the Java compiler and JVM, and
> > only be able to run those Java programs that you yourself will write
> > and compile on your compiler?
> 
> s/write and// - there's no reason to assume that using VAX floats means
> that you will _write_ all your own code.  Even aside from that, there's
> plenty of java code - even bytecode - that doesn't even use floats.

Okay, in the case that the code don't use fp, you will survive, I'll give
you that. Hmmm, I wonder if there is any statistics on how often people
use fp nowadays...?
And yes, you can of course compile other peoples source code as well.

> > Why on earth in that case?  C++ should work just as well for you.
> 
> You consider C++ equivalent to Java if you're working with source?  I
> sure don't.

And why not? Java have more limitations for sure, but I don't think that
is a problem in this case.

The big point with Java is portability, and if you skip that, then I don't
see a point in Java at all.

	Johnny

Johnny Billquist                  || "I'm on a bus
                                  ||  on a psychedelic trip
email: bqt@update.uu.se           ||  Reading murder books
pdp is alive!                     ||  tryin' to stay hip" - B. Idol