Open fredrik-johansson opened 2 years ago
This essentially calls mpfr_round_p, which does not have a special case for checking when we can round to nearest. For example, 1.0 +/ eps can clearly be rounded to nearest when eps is small enough, but this can't be detected with mpfr_round_p.
This essentially calls mpfr_round_p, which does not have a special case for checking when we can round to nearest. For example, 1.0 +/ eps can clearly be rounded to nearest when eps is small enough, but this can't be detected with mpfr_round_p.