[PLUG] Brain the size of a planet

M. Edward (Ed) Borasky znmeb at cesmail.net
Wed May 4 14:59:37 UTC 2005


Yeah ... I know ... I've spent more of my programming time over the
years working in FORTRAN than any other language. I used to refer to it
as the Listerine of programming languages -- "I hate it but I use it". I
spent ten years at Floating Point Systems -- how could I *not* love
FORTRAN? :)

Have you looked at R? You can call FORTRAN, C and C++ code from R (and
vice versa, IIRC) and nearly every piece of useful numerical and
statistical knowledge exists as a package. There is a GRASS interface
and I believe there is also a map database. I use it where I work to
analyze computer performance statistics. It's also widely used in
biology and finance. I even know someone (me) who is using it in
algorithmic composition and synthesis of music.

A few years ago, I was looking for an open source math package to
integrate into a computer performance analysis package. I looked at Perl
Data Language (PDL), XLisp-Stat, Octave (a Matlab look-alike), SciLab
(another Matlab look-alike) and R. R won, hands down, and that was when
R was quite young -- I believe the release was 1.2 or maybe earlier. R
is now at 2.1, and I don't think any of my other choices have evolved
much at all since then. I think I bet on the right horse. :)

This discussion in various forms shows up all over the Open Source
community -- start something new and really cool, or contribute to
something that's useful but already exists? The short answer, I think,
is that we need both. I've clearly cast my lot in the second camp, as
the Gentoo and R mailing lists will show.

Rich Shepard wrote:

> On Wed, 4 May 2005, M. Edward (Ed) Borasky wrote:
>
>> Bah! As a recovering FORTRAN programmer, I can only say that what
>> high-performance computing exactly *doesn't* need is another language!
>
>
> Ed,
>
>   Much as we might wish for Fortran to go away, it's still alive and
> well in
> the ecological modeling world. A very powerful basin model (that
> integrates
> very well with GRASS) is the Soil and Water Assessment Tool (SWAT)
> created
> and under continual development by the USDA's Ag. Research Station in
> Temple,
> TX. SWAT is one of the few EPA-approved basin models that does hydrology,
> erosion, sediment transport and all sorts of other useful stuff. And,
> it's
> still written in Fortran.
>
>   A lot of the US Geological Survey code is still Fortran (and still
> without
> comments) and so are geodetic models.
>
>   They all build nicely with gcc.
>
> Rich
>



More information about the PLUG mailing list