Errol: An always correct method for printing floating-point numbers [pdf]cseweb.ucsd.edu1 pointfanf26 years ago