REXX can handle floats, but it isn't terribly suited to numerical work. It lacks mildly interesting mathematical functions, may that be trignometric or otherwise. So I checked out rexxmathlib.
When I downloaded rexxmathlib, there were worrying comments about round-off error in the documentation. To give you an idea of what this means, I tried solving a simplified version of this problem in DICE. It was simplified so that I could find an exact solution and compare it to an iterative solution. The difference between the two solutions was about 1%. I would expect REXX to be greater.
The rexxmathlib documentation also suggested that floats were stored in 15.9 format. While it wasn't a problem in this case, I often deal with problems which have very big or very small numbers. For that, I need floats, and not some sort of fixed precision decimal (assuming I have the right name).
I tried DICE. It was relatively small (can you tell that I have dealt with GCC too much :pissed: ), and I was impressed with the speed on my 68030, which is actually behaving well for the moment. :-D
But I'll try Python later because I think it has arbitrary precision arithmetic libraries. Or was that Perl. I forget. At any rate, while DICE is fast I have found compilers on other platforms to be sluggish. Why is beyond me. So interpreted langages can be nice.
