math programming

math programming in sepples is rather complicated to the point that books are written on the subject. first there is the factor of limited precision of types, like how much a double or long double can hold. so is it preferable to use dynamic programming languages like Python and Perl that dont have any limit on length of floating point precision? and is it true that twos compliment creates gaps in the number line? do calculators use twos compliment? I noticed some high end calculators use the old Z80 or 6502 CPUs, but is there added arithmetic circuitry or can you rely on these standard CPUs to provide sufficient math processing?

An array of chars, with an integer offset for decimal point.
Or, two char arrays: one for the real part of the number and one for the decimal.

If you really need big numbers, arrays of unsigned long ints where each element contains the maximum amount and then starts adding to the "next" element.