This is the mail archive of the
gsl-discuss@sources.redhat.com
mailing list for the GSL project.
Re: Java wrappers
- From: Luc Maisonobe <Luc dot Maisonobe at c-s dot fr>
- To: Nigel Wetters <Nigel dot Wetters at lshtm dot ac dot uk>
- Cc: gsl-discuss at sources dot redhat dot com
- Date: Mon, 17 Mar 2003 13:28:03 +0100
- Subject: Re: Java wrappers
- References: <se75a06f.063@s-webmail.lshtm.ac.uk>
Nigel Wetters wrote:
> Probably heresy to suggest using an interpreted language for scientific
> programming, but has anyone produced either Java wrappers around the
> GSL, or Java implementations?
Not an heresy at all, and I don't feel Java can be qualified as
interpreted anymore: hotspot has been round since years now and it works
great. There are several studies showing that Java can be as fast as C++
or fortran in many cases, even including large size numerical problems.
Of course, it is very easy to write bad Java programs and have extremely
inefficient programs. The main thing to avoid (as always with object
oriented languages) is to avoid rebuilding a tremendous amount of small
temporary objects over and over as this is the main reason for object
overcost.
I have written a small library of mathematical algorithms for my own
purposes and for use in my office (space dynamics). It is really far
from the completeness of GSL, but perhaps you can find there what you
want (or ask me for more). You can find it there :
http://www.spaceroots.org/archive.htm#MantissaSoftware
Luc