PawCalc uses IEEE 754 double precision numbers. This gives PawCalc a precision of around 14 digits, and exponents in the range -300 to +300.
Take care to stay within these limits. This also applies to intermediate calculations such as the implicit conversion to and from base units in unit conversions. For example, to convert from parsec to lightyears, it is enough to multiply by approximately 3.26. However, PawCalc converts from parsec to meters and then from meters to lightyears resulting in multiplication by 3.08e16 and then division by 9.46e15.
Not all real numbers have an IEEE 754 representation. Instead such real numbers are approximated by the nearest IEEE 754 number. Such approximations lead to small errors. For example, the decimal representation of the fraction 1/7 is 0.142857142857... where 142857 repeats forever. Try calculating '1/7-0.142857142857'. The result is 1.42857948e-13. The last three digits are 948 and not 142.
In some cases, the trig functions, the error functions, and other functions return results that are slightly off. For example, in degree mode using scientific viewing, 'sin 360' is not 0 but -1.133e-15. However, with normal viewing, the result of 'sin 360' shows as 0.
Expressions are evaluated from left to right. Because of the limited precision, 1+1e-10-1 is not equal 1-1+1e-10. This is a well-known problem in numerical analysis. There are a few rules of thumb for avoiding such problems:
Never subtract almost equal numbers and be careful when comparing such numbers.