Am Mon, Nov 22, 2021 at 04:10:15PM +0100 schrieb Giulio Paci: > The failing test case relies on the assumptions that, when a and b have the > same double value: > 2) "a - b" is 0.0.
(comment only based on this mail, did not check the code) usually (C,C++) you cannot reliably say "if ( (a - b) == 0 )" or "if (a == b)" as floating point cannot always represent the numbers exactly enough. Usually comparing floats for almost-equality is done, something like if (fabs(a - b) <= FLT_EPSILON) (but FLT_EPSILION (or DBL_EPSILON) might not always be suitable, for example it depends how much rounding you might have picked up on the way -- tobi
