Getting surprising behavior when casting

I'm trying to cast a float to an unsigned int and getting some surprising behavior.

Code:

---

float x = -1.0;
unsigned int y = (unsigned int)x;
printf("y = %d\n", y);

---

The output of this code changes depending on which compiler I use. Sometimes I get -1 and sometimes I get 0. Not really sure why though.

edit: Tried searching before posting but tried a different search term after posting and found a stackoverflow thread explaining why this is not portable. A bit surprising, but good to know.

What was the problem?

The unsigned cast of a float. A signed cast works ok.

If you want to print an unsigned int then you should use "%u" and not "%d". Maybe that's your problem.

Regards.

Anyone care to expand on this problem, i.e how negative integers/floats are cast into postive integers/floats? I feel there is a topic here..

O_o

There really isn't much to explain. Well at least, there isn't much to explain in the direction you imagine.

The compiler simply decided that, because an `unsigned' value can't be negative, the assignment could be skipped.

As you noted, using a `signed' integer "works". Using code that trumps the optimizer--such as reading the value from `scanf'--also works.

Oddly, the user gemera recently reference the issue in the other direction.

Yep. The behavior completely conforms to the standard.

Soma