[Date Index] [Thread Index] [Author Index]
Re: GUI for Mma
function: Mathematics. The kernel is very good, if a little slow, but both the 2.2 and 3.0 front ends get in the way of the language. The 3.0 front end has many advertised features that simply don't work.(file conversions mostly) The 2.2 front end editor didn't even have the functionality of the simple DOS editor. I think Wolfram ought to get out of the front end business and give us "The Mathematica C++ Library" which would be a bunch of c header files that could be loaded in to incorporate the features of the language desired for a particular applicataion. This would immediately remedy the two most serious limitations to mathematica: No stand-alone executables and non-portable code. I use mathematica heavily now during my PhD program, but when I return to my Laser research job, I realize that I can only use mathematica for my own enjoyment or for numerical speculation. I can never specify code for a contract to be written in mathematica. Were there such a thing as a "Mathematica C++ Library", mathematica could immediately enter use in the research and industrial community in general. I like mathematica better than writing C code because I don't have to be burdened with memory allocation, remembering all those pointers and whether I am passing by reference or value and I don't have to write my own graphing codes. Also, as a general engineering principle, the more things something does, the poorer it does any one of its tasks. Two mathematica examples of this are: NIntegrate: it is so robust and general that it is abyssmally slow, so I only use it to check my work and nearly always write a specific integration routine appropriate for the problem at hand. 3.0 kernel: It does so much more than the 2.2 kernel that it is much slower according to my experience and independent review. Anyway, the more functions you try to add to mathematica, the slower it is going to be and will get in the way of the primary reason anyone uses it: Fast and accurate answers.