|
|
|
|
|
|
by nwhitehead
3877 days ago
|
|
|
This might be true on really weird platforms, but with IEEE-754 floating point this isn't true. Integers within a big range are exactly representable without any rounding error. Doing floor((float(13) + float(13))/2.0) should be exactly 13. It is entirely possible that the weird platform was supposed to be using IEEE-754 arithmetic but had a buggy math library (depressingly common). |
|
|