A simple "fuzzy" tolerance will not solve the problem, only mask
it for the most common cases.  A single floating point addition
of two numbers can result in ZERO bits of precision.

Reply via email to