has basic floating point support with standard linear math operators , but lacks math functions present in most PC based compilers. 
This is obviously planned for in the dsPIC compiler, however the hope would be that the next release would provide these functions (the Michrochip library) 
, not only provide the same libraries as the version for PIC. I think the compiler will need something in addtion to the PIC version to make it attractive beyond the speed increase.Personally I would be interested in this for instrumentation that would require non-linear math expressions, which would be really hard to do with integer math. Examples would be exp, ln, sqr, exp10, log, polynomials et., etc.
Fourier transforms etc. are also in a free Microchip library; however the basic math library should come first.
Also needed for floating point is better utilities for conversion to other types, for instance to provide output back to a D/A converter.
A third thing would be conversion to stings with formatting (see separate post).
--
OT
