long ----- float

[Geoffrey Vlassaks]

Hi,

long l = 1L;

float f = 1.0F;

//insert code here

f = l; //compiles fine
l = f; //gives error
I don't understand it, because long is 64-bit and float is 32-bit
Thx,

[Jong Limb]

It's not a matter of bits; it's a matter of precision. Longs can only hold integral values. Floats can hold fractional values. If your float variable contained the value 12.34, what happens to the 0.34? It gets truncated. The compiler will flag this for you as a loss of precision.

[Corey McGlone]

We often refer to things as widening or narrowing conversions. A widening conversion is a conversion that is "safe" - there can be no loss of data in a widening conversion. A narrowing conversion, however, is not safe. You can lose data in a narrowing conversion.
To consider something a widening conversion, it must pass one test - all values representable by the first data type must be representable by the second. If that is possible, it is considered a widening conversion.
Based on that, it's easier to see that a conversion is considered a widening or narrowing conversion based on the ranges of the data types, not the number of bytes those data types contain. If the range of the original data type is a subset of the range of the conversion data type, the conversion is considered to be widening. If not, the conversion is a narrowing conversion.
As every value representable by a long is also representable by a float, converting a long to a float is a widening conversion. As every float, however, is not representable by a long, converting from a float to a long is considered a narrowing conversion.
I hope that helps,
Corey