The OP was using java for a _new_ project, and I was responding favorably to
that, not with any implication that octave needs to be ported.
I guess it isn't hard to understand that the scilab team prefers java to fortran. Other comparisons are not so clear-cut.
However I think there is a clear demand for octave functionality outside the boundaries of posix and X, as well as for improved graphics compatibility with matlab. Appropriate tools helping meet that demand should be looked at closely.
----- Original Message ----
From: A. Scottedward Hodel <address@hidden>
To: address@hidden
Sent: Wednesday, November 15, 2006 1:31:52 PM
Subject: Re: matlab,octave clone in java. JMathLib
I agree for another reason, which is that the C++ octave interfaces and numerical libraries provide a useful and straightforward way to interface octave with external legacy code. This is useful in many
environments where FORTRAN and C++ code is the preferred language of development.
At about a dozen years old, octave is, in some sense, becoming legacy code in the sense that we've got a huge base of software that works. Conversion to another language is nowhere near a trivial effort, so a case would be need to be made that it needs done and, if such a case is made, a large group of people would need to volunteer to do it.
Just my 2 cents worth of opinion.
Yours,
On Nov 15, 2006, at 9:38 AM, Quentin Spencer wrote:
James R. Phillips wrote:
The current scilab roadmap at
www.scilab.org shows that version 5 is to
be coded primarily in java. Not due for about a year though - nothing
available to download yet that I can see.
Java is looking like possibly a good choice for open source
math
software, with Sun moving toward gpl2 licensing for it. The scilab
license will be less free than the java it is coded in! I hope that
may change in the future.
I don't know whether underlying fortran libraries like lapack or atlas can be linked with it.