Questions

1. A bytecode interpreter could use an instruction set with up to three addresses (operands) per instruction, such as three-address code. Instead, the Jzero interpreter uses zero or one operands per instruction. What are the pros and cons of using three-address code in the bytecode interpreter, such as in intermediate code?
2. On real CPUs and in many C-based bytecode interpreters, bytecode addresses are represented by literal machine addresses. However, the bytecode interpreters that were shown in this chapter implement bytecode addresses as positions or offsets within allocated blocks of memory. Is a programming language that does not have a pointer data type at a fatal disadvantage in implementing a bytecode interpreter, compared to a language that does support pointer data types?
3. If code is represented in memory as an immutable string value, what constraints does that impose on the implementation of a bytecode interpreter?