BrewlearnWidening conversions in Java
Automatic safe conversions (int → long → float → double) and precision loss in long → float.
Small into big: automatic
Java converts a value to a wider type automatically, no cast needed. This is a widening conversion. Along the chain byte → short → int → long → float → double (and char → int), the value never overflows.
int i = 7;
long l = i; // int to long
double d = i; // int to double
int c = 'A'; // char to int: 65Your turn
What does this print?
int apples = 3;
double d = apples;
System.out.println(d);33.03.00
Show the answer
The int 3 is widened to the double 3.0. Printing a double always shows at least one digit after the point.
Big into small needs a cast
The opposite direction is narrowing and could lose data, so Java demands an explicit cast. If someLong is a long, int i = someLong; doesn't compile. (Casts are the next lesson.)
int i = someLong; // error
int j = (int) someLong; // ok, a castRange vs precision
float and double cover enormous ranges but keep only a limited number of significant digits: float about 7 (24 bits of precision), double about 15–16 (53 bits). Widening keeps the size of the number, but low digits may be rounded.
Your turn
A long goes into a float and back. What comes out?
long n = 123_456_789L;
float f = n;
System.out.println((long) f);123456789123456792Compile error
Show the answer
long → float is legal widening, but float has only 24 bits of precision. 123,456,789 can't be stored exactly, so it's rounded to the nearest float: 123456792.
long → double isn't lossless either
double has 53 bits of precision, while long has 64. Longs above 2⁵³ (about 9 quadrillion) can be rounded when widened to double, silently.
long big = 9_007_199_254_740_993L;
double d = big;
// (long) d is 9007199254740992In real projects
IDs are often 64-bit longs. Push them through a double (some JSON parsers, or JavaScript, where every number is a double) and large IDs silently change. Twitter's API added string IDs (id_str) for exactly this reason.
Key takeaways
- Widening happens automatically, with no cast
- int → long → float → double never overflows
- long → float / double can round off low digits
- float keeps ~7 significant digits, double ~15–16
16,777,217 (2²⁴ + 1) is the smallest positive whole number a float can't store exactly. Every integer up to 2²⁴ fits perfectly.
Practice questions
What does this print?
int i = 7;
double d = i;
System.out.println(d);- 7
- 7.00
- 7.0
- Compile error
Check your answer
7.0. The int is widened to the double 7.0. Printing a double always shows at least one digit after the decimal point.
Which assignment needs an explicit cast? (someLong is a long, someInt an int, someFloat a float)
- long l = someInt;
- int i = someLong;
- double d = someFloat;
- float f = someLong;
Check your answer
int i = someLong;. long → int is narrowing and could lose data, so you must write (int). The others are widening conversions that happen automatically.