When is a float variable not equal to itself

Question

I was asked this questino on a programming test. The question was, I was passed in a float as a parameter, and asked when would the following code be false.

bool result  = (floatValue == floatValue);

I couldn't think of a valid reason or a situation of when this would be false and still can't. In the end, I answered that there will never be a case when this would be false. Was wondering if anyone can give me some example(s) of when this would be false

http://stackoverflow.com/questions/18661635/why-two-float-type-variables-have-different-values — Ritesh, Oct 02 '15 at 03:54
@M.M how comes? Don't RHS and LHS have exactly the same floating point representation? — vsoftco, Oct 02 '15 at 04:08
@vsoftco one could be in a register with higher precision as a result of previous calculation — M.M, Oct 02 '15 at 04:32
@M.M That's true when comparing two different float values, but in this case I think you're comparing one variable with itself. — 1201ProgramAlarm, Oct 02 '15 at 04:37
@M.M. not sure I understand. `floatValue` here refers to the *same* lvalue. It's stored only once. — vsoftco, Oct 02 '15 at 04:38
The compiler could use a copy left in a register from an earlier calculation for one of the operands — M.M, Oct 02 '15 at 05:08
BTW you mean "same object", not "same lvalue". An lvalue is a category of expression. — M.M, Oct 02 '15 at 05:08

score 12 · Accepted Answer · answered Oct 02 '15 at 03:51

12

floatValue == NaN

If floatValue is the result of a computation that is Not a Number, like 0/0.

answered Oct 02 '15 at 03:51

1201ProgramAlarm

32,384
7
42
56

OH. I was totally forgetting NaN. Does this also apply if floatValue is INF? Or is this just for when floatValue is NaN? – dwnenr Oct 02 '15 at 03:55
Only for NaNs. Infinities compare normally (-infinity < any real number < +infinity; +infinity == +infinity; etc.). – 1201ProgramAlarm Oct 02 '15 at 04:35

When is a float variable not equal to itself

1 Answers1

Linked