Re: The Plan

["Martin P. Pokorny" <mpokorny@glatmos.com>]

Hello,

At 11:19 AM 8/23/1999 -0600, Mark Galassi wrote:
>Could y'all please all post a note stating whether you use *just C*,
>or *just C++*, or *both languages* when you use somebody else's
>numerical routines (like GSL) in your own scientific programming.
>
>I think it's important to state what you have written yourself, and
>what you have used so far, rather than what you might use in the
>future.

I'm a relatively new user of GSL, but I have a great interest in its
future, and thought I'd put in my two cents worth regarding the present
discussion of "The Plan". Most of the scientific software I have written
was done in C. I have a bit of experience using C++, but I prefer to avoid
it when possible (other OOPLs and VHLLs are more to my tastes for a variety
of reasons).

Recently I have been working on an Eiffel binding to GSL and other
numerical computing libraries. I prefer "external" libraries with a C API
because, for interfacing with external routines, the only method supported
by all the (Eiffel) compilers is via a C interface. Furthermore, class
design in Eiffel, and other OOPLs as well (e.g., Java), may not be
"homeomorphic" to C++ class design in all cases, imposing further
difficulties for other language users. Thus I much prefer an
"Object-oriented library written in C," in Mark's words. I agree with Mark
that GTK+ is a good example of that type of design. In my experience, using
a VHLL on top of a library with a C API is a simple and flexible method by
which to use a library.

Happy trails,
  Martin

-------------------------------------------------------
Martin P. Pokorny           voice: 520.806.7300
Global Atmospherics, Inc.     fax: 520.741.2848
2705 East Medina Road      e-mail: mpokorny@glatmos.com
Tucson, AZ  85706-7155       http://www.glatmos.com