Why is the order of downcasting in Java float <- long?

Question

My textbook says this:

Converting from a larger type to a smaller type is called a down cast, also known as range narrowing.

Byte <- short <- int <- float <- long <- double

Why is long higher than float on the downcasting/narrowing list?

Isn't the range of values represented by a float much wider than the range represented by a long?

E.g., when casting types (assuming f represents a float and l represents a long), you can do do f = l, but this requires an explicit cast: l = (long) f. Doesn't this show that a float should be higher than a long on the downcasting list?

[Why does Java implicitly (without cast) convert a `long` to a `float`?](http://stackoverflow.com/questions/1293819/why-does-java-implicitly-without-cast-convert-a-long-to-a-float) — Sotirios Delimanolis, Aug 31 '16 at 23:24

score 0 · Accepted Answer · edited Sep 01 '16 at 00:40

A long and a double are both 64 bits whereas a float is only a 32 bit data type. Thus it takes a downcast to turn the long's 64 bit number into a 32 bit.

When programming, float has a wider range of number, but the long uses its 64 bits to create a more precise number. This is the reason you use float when you need to save space on a less precise number, but for something where precision is important (i.e., currency, scientific calculations, etc) a long is what you should be using.

If you're curious you can check out the Java Doc's here

Why is the order of downcasting in Java float <- long?

1 Answers1