The result is correct according to C type propagation rules and will be obtained with any C compiler. 100/1000 is calculated as integer division (result 0) and then converted to float.
You may want to write e.g.:
Code:
float32 a = 100.0/1000;
float32 a = 100/1000.0;
float32 a = float(100)/1000;
In any case, at least one of the numbers must be of the float type to cause a float division.