Types in programming language:

• specify length of data (how many bits)
• help prevent errors
• enforce program invariants

In assembly and C:

• possible to case integer into a pointer
• enables program to write to random memory locations

Types in JVM

• as in (early version of) Java (sometimes simpler types)
• goal: JVM byte code verifier ensures memory safety
  • cannot overwrite memory locations
  • can avoid doing certain checks while running JVM

One difference Java vs Byte codes:

• java treats boolean as int
• they are implicitly converted into integers primitive conversions
• and they are implicitly converted back from integers :)

Little support for boolean operations

• intermediate boolean values are often invisible (encoded in program points)
• can manipulate them as integers

Using cafebabe, our wonderful bytecode generation library, we generate this class:

public boolean test(int, boolean);
  Code:
   0:   iload_1
   1:   iload_2
   2:   bipush  37
   4:   iadd
   5:   iadd
   6:   ireturn

public static void main(java.lang.String[]);
  Code:
   0:   getstatic       #19; //Field java/lang/System.out:Ljava/io/PrintStream;
   3:   new     #3; //class HW
   6:   dup
   7:   invokespecial   #20; //Method "<init>":()V
*  10:  iconst_3
*  11:  iconst_2
*  12:  invokevirtual   #22; //Method test:(IZ)Z
   15:  invokevirtual   #28; //Method java/io/PrintStream.println:(I)V
   18:  return

}

Using

java HW

in Java HotSpot(TM) Server VM (build 1.6.0_03-b05, mixed mode), this:

• successfully passes bytecode verifier
• happily returns 'the' answer

We adopt the following convention in code generation for JVM

The generated code will use 'int' to represent boolean values in

• local variables
• parameters
• intermediate stack values (if needed–often not needed)

In such cases, the 'int' variables will always be either

• 0, representing false
• 1, representing true

Other conventions are possible