> lemme try this again. You can in no way rely on the last digits of a float > output. Those depend on your system, your CPU, your ini settings and any > operation you do with floats reduces the accuracy....
It's still a regression, and should be documented. S
