Follow-up from "Fix/real comparison"
The following discussions from !1134 (merged) should be addressed:
-
@ccantwel started a discussion: (+1 comment) Is it necessary to restrict this to cases where the operands are the same type?
Presumably if I compare a
float
and adouble
, the relative tolerance for equality would be determined by the most imprecise data type? -
@ccantwel started a discussion: (+1 comment) Could get numeric limit for both T1 and T2 and use the smallest?