Java byte to int Coercion

Question

I have a question about the code below. The code is taken from my Programming Languages book.

byte x, y, z;
...
/* The values of y and z are coerced into int and int addition is performed */
/* The sum is converted into byte */
x = y + z;

My question is why Java does a coercion like that. Do you have any ideas?

Thanks in advance.

It's because that's how the JVM works (it can't do addition on bytes): http://stackoverflow.com/questions/18483470/is-addition-of-byte-converts-to-int-is-because-of-java-language-rules-or-because — wkl, Mar 03 '14 at 17:31
I understand that y and z become ints during addition, but doesn't setting this back to x cause a compile error without a cast? — martinez314, Mar 03 '14 at 17:38
This page contains very useful information regarding this issue: https://www.artima.com/underthehood/bytecodeP.html — Hessam, Mar 03 '14 at 17:46

ikegami · Answer 1 · 2014-03-03T17:38:45.290

2

To keep the JVM instruction set simple, it doesn't implement an addition operator for each integral type. Specifically, it can't add bytes. (See the list of JVM ops) This makes the JVM simpler, and probably more portable.

edited Mar 03 '14 at 17:38

answered Mar 03 '14 at 17:32

ikegami

367,544
15
269
518

and what about longs (onx64) ? – Svetlin Zarev Mar 03 '14 at 17:37
@Svetlin Zarev, Misspoke. Fixed. – ikegami Mar 03 '14 at 17:39
1

@SvetlinZarev there is an instruction for adding longs (ladd), which pops four 32bit values (2 for each number) from the stack, adds them, and pushes two 32 bit values onto the stack again. – maxdev Mar 03 '14 at 17:40
@ikegami Nice to see you on Java tag :) You can add this [JVM Spec](http://docs.oracle.com/javase/specs/jvms/se7/html/jvms-2.html#jvms-2.11.1) link to your answer though. – Rohit Jain Mar 03 '14 at 17:41

score 1 · Answer 2 · answered Mar 03 '14 at 17:35

This mirrors the java byte code, the jvm, java virtual machine, that in the specification uses an int to store a single byte variable on the stack. It uses iadd for adding two bytes. See the jvm instruction set.

One might argue that it is a bit of overspecification: mentioning an implementation choice of the reference implementation.

maxdev · Accepted Answer · 2014-03-03T17:44:16.447

1

In the JVM, every stack element has a size of 32 bits. The actual addition works like this:

The two bytes are pushed on the stack as 32bit values (therefore they are int)
The iadd instruction is called, which pops two values from the stack and adds them
The resulting integer is pushed on the stack again

This is why you have to cast the resulting value (of type int) to a byte again.

edited Mar 03 '14 at 17:44

answered Mar 03 '14 at 17:38

maxdev

2,491
1
25
50

Java byte to int Coercion

3 Answers3