More specifically, you mean in an integer numeric field. If you use a floating point field, it will work correctly.
When a Panorama 6 function or operator has operands with different data types, it converts both of them to the final field type, then performs the operation. In this case
is an integer, and
is a floating point number from 0 to 1. Since an integer and a floating point number are different types, Panorama converts them to the type of the integer if the result is going to be an integer. Then it does the multiply.
If you convert the output of rnd( to an integer, you will get 0 or 1, so that is what you are seeing.
If the final result is a text field, Panorama treats that the same as a floating point field, so in that case 2345 is converted to a floating point number, and rnd( is already a floating point number. So it works great.
Why does Panorama 6 do this in such a dumb way? Remember, this program was originally started in 1986, when floating point operations were very, very slow. So converting integers and fixed point values to floating point, then back again, would be quite slow.
You can make your formula work in an integer field by explicitly converting everything to float(, leaving Panorama with no freedom to decide how to do the conversion.
Panorama X doesn't have this problem. If operands don't match, they are always converted to floating point. And the fixed point numeric types don't even exist any more. Floating point is super fast now, so that isn't a problem, and you no longer have to worry about underflows and overflows any more.