Thread Previous • Date Previous • Date Next • Thread Next |
It seems that the 'official' way of rounding BIUs is something like: fval < 0 ? fval - 0.5 : fval + 0.5 i.e. 'away from zero', since the double-to-int C is a truncation (i.e. 'toward to zero'')... what is the rationale for this? is this to preserve, for example, the usual symmetry of packages? -- Lorenzo Marcantonio Logos Srl
Thread Previous • Date Previous • Date Next • Thread Next |