On Thu, 10 Dec 2015 09:41:21 -0800 Evan Gates <evan.ga...@gmail.com> wrote:
Hey Evan, > The problem with that is bc can't use a normal bignum library and be > POSIX compliant as all the internal math is required to be in base 10. > "Internal computations shall be conducted as if in decimal, regardless > of the input and output bases, to the specified number of decimal > digits."[0] That's why implementations of bc tend to have their own > arbitrary precision stuff built in. What does the base have to do with the results? > That being said a suckless bignum library can still be very useful, > and sbase tends to stray from POSIX in a few places. This may be one > such place that makes sense to ignore POSIX. For bc we need simple > arithmetic, sqrt, integer exponents, sine, cosine, arctangent, natural > log, exponential function, Bessel function of integer order, and a way > to work with the bc concept of scale.[0] Depending on your performance > desires many those can either be implemented in the library or in bc > using a simpler library. > One last off topic question, should sbase have a bc that calculates, > or just translates to dc (which is not a POSIX tool, but probably > useful to have in sbase). To be honest, I'd go for implementing dc(1) and using the normal bc(1) based on dc(1). RPN is also easier to implement. Cheers FRIGN -- FRIGN <d...@frign.de>
