Though it is not possible in Java, the JVM can have multiple methods with the same name and parameter types so long as they have different return types. The JVM spec seems to call it "method descriptor" but I called it signature as I think that's more attention grabbing (and I'm showing how the JVM spec and Java spec differ).
No two methods in one class file may have the same name and descriptor
You can experimentally test this yourself with the Jasmin project. Jasmin is a tool that lets you write assembly code to assemble to JVM byte code. I also used this classfileanalyzer project to get some base Jasmin assembly.
For the lazy, here is the Java code I started with,
public class Example {
public static void main(String[] args) {
System.out.println(foo());
System.out.println(fooInt());
}
static String foo() {
return "1";
}
static int fooInt() {
return 1;
}
}
Then I got this Jasmin code and replaced each fooInt with foo
After assembling Example.class with Jasmin and running javap -c Example I get this output,
Compiled from "Example.java"
public class Example {
public Example();
Code:
0: aload_0
1: invokespecial #28 // Method java/lang/Object."<init>":()V
4: return
public static void main(java.lang.String[]);
Code:
0: getstatic #33 // Field java/lang/System.out:Ljava/io/PrintStream;
3: invokestatic #32 // Method foo:()Ljava/lang/String;
6: invokevirtual #10 // Method java/io/PrintStream.println:(Ljava/lang/String;)V
9: getstatic #33 // Field java/lang/System.out:Ljava/io/PrintStream;
12: invokestatic #11 // Method foo:()I
15: invokevirtual #24 // Method java/io/PrintStream.println:(I)V
18: return
static java.lang.String foo();
Code:
0: ldc #27 // String 1
2: areturn
static int foo();
Code:
0: iconst_1
1: ireturn
}
And of course, running Example.class outputs
1
1
I hope you've enjoyed this interesting quirk of the JVM. It is a good reminder that while the JVM is primarily for Java, there are other languages that use it and some might even use this in a meaningful way somehow.
Has there ever been official discussion regarding why you can't do this in Java, if it's a thing the JVM allows? Maybe there are performance issues with code analysis and/or compilation?
My gut feeling is the answer will probably be "because C does it like that" but I doubt there's a primary source saying that anywhere.
I suppose one tricky thing with object oriented code might be deciding which method to use. The naive solution is just use the most specific but what if there are equally specific ones? Say in the example above I used Integer instead of int and ran this,
Object o = foo();
Both String and Integer are as "specific" as each other.