Re: Proposal to get Stephen to Improve the lot ofSpace Shuttle Programmers

*To*: mathgroup at smc.vnet.net*Subject*: [mg60488] Re: Proposal to get Stephen to Improve the lot ofSpace Shuttle Programmers*From*: Bill Rowe <readnewsciv at earthlink.net>*Date*: Sun, 18 Sep 2005 01:15:45 -0400 (EDT)*Sender*: owner-wri-mathgroup at wolfram.com

On 9/16/05 at 3:51 AM, beliavsky at aol.com wrote: >Bill Rowe wrote: >>On 9/14/05 at 3:27 AM, davidannetts at aapt.net.au (David Annetts) >>wrote: >>>>Can I get a conversation started about this topic? Will a tool >>>>to convert FORTRAN to mathematica code be useful? >>>I'd suggest that a tool going the other way (ie Mathematica to >>>F95, F2K or some variety of c++) would be far more useful than a >>>tool to convert Fortran code to Mathematica. >MathCode F90 >http://www.wolfram.com/products/applications/mathcodef90/ converts >Mathematica to Fortran 90 . As others have pointed out, this is a rather expensive tool and does not address the original poster's suggestion. >>There is an obvious use for such a tool. There exists quite a bit of >>useful FORTRAN code in various texts and other places. And I for one, >>do not have ready access to a FORTRAN compiler. >There exist several free Fortran compilers for Fortran 77 on >Windows, Linux, and other platforms, and two Fortran 95 compilers, >g95 and gfortran. A Fortran 95 compiler can compile most old >Fortran code written in Fortran 77 or Fortra IV. See >http://www.dmoz.org/Computers/Programming/Languages/Fortran/ >Compilers/ for links to compilers. Thanks for the useful suggestion, but it really doesn't invalidate the point I was trying to make. I am certain I am not the only person using Mathematica that is far more competent with Mathematica than FORTRAN. So, the ability to translate FORTRAN code into Mathematica code reasonably automatically would be useful. Yes, the resulting code will be slower and will likely be less than optimum Mathematica code. But this observation does not change the usefulnees of being able to translate FORTRAN to Mathematica. In fact, the real time it takes to solve a particular problem consists of the execution time plus the time it takes me to get the computer setup to solve the problem. And this latter time is often has far greater impact than differences in execution time. I've no need to stare at a computer screen while the computer is solving my problem. Assuming that is a significant time, I can use it to accomplish something else. OTOH, I really cannot use the time I spent creating a program or debugging a program to do something else. The point is focusing on the improved execution speed of compiled code versus Mathematica may well be focusing on the least important aspect of getting a particular problem solved. -- To reply via email subtract one hundred and four