Two-Way Translation & Symbolic Addressing

1. Translation Recap — Compiler vs Interpreter

Translation = turning your source code into something the CPU understands (machine code = the language the computer can understand). Two ways:

(recap row facts are the standard textbook ones from System Software; the recording only points back to them.)

2. Partial Compilation + Partial Interpretation

Idea: do half of the translation by compiling, then do the other half by interpreting ⇒ we combine the advantages of compilation and of interpretation.

3. Example — How Java Source Code is Translated

Java is the language that uses the two-way process. Past paper: “Describe how a Java source code program is translated.”

What is byte code? (sir’s link to assembly)

Remember the assembly questions: LDM #97 → you wrote the binary op-code of LDM from the table, then converted 97 to binary in a second box. That pair of boxes is exactly the idea of a byte code: instruction + the data to work on, written in 0s and 1s.

(LDM op-code from this paper’s table; 97 = 64 + 32 + 1 = 0110 0001. Used only as the analogy — real Java byte code has its own op-codes.)

The Java compiler makes this half-way (“intermediate”) code; it is not complete machine code, so nothing is run yet.

What is the JVM?

A virtual machine = a software “device” that behaves like a machine. Like running another OS on a normal PC, the JVM lets many different CPUs / PCs each interpret the same byte code. So: byte code → interpreted through the Java Virtual Machine.

4. Why Partially Compile & Partially Interpret?

Advantage ① — runs on ANY platform, no re-compiling

• Normal full compilation on Windows ⇒ complete machine code for that CPU/OS only. It will not work on Linux or iOS (it works only where it was compiled).
• Byte code is just numbers / bits (0s and 1s) ⇒ platform-neutral ⇒ can be used on any platform.
• The same Java program runs on Windows, Mac or Linux without re-compiling (no “compile on Windows, then compile again for Linux”). Compile once to byte code and you are done.

Advantage ② — optimised for each CPU

• The byte code is optimised for that particular CPU (whatever OS / hardware is installed) during the interpretation phase.
• The virtual machine can use information about the hardware architecture: number of cores, memory system / RAM, memory consumption, storage — different for every CPU.
• It uses this to generate the most efficient machine code = the code that makes the CPU run fastest.
• So the same byte code executes on different CPUs, but the machine code is generated at run time for that particular CPU ⇒ optimised.

5. Instruction Set & Addressing Modes (Board)

Processor has one general-purpose register — the Accumulator (ACC) — and an Index Register (IX). Part of its instruction set:

Each instruction and each data line takes one memory location ⇒ one line = one address (add 1 per line).

6. Two-Pass Assembler & Symbol Table

An assembler translates assembly language into machine code. A two-pass assembler reads the program twice because a label can be used before it is defined (e.g. JMP Finish comes before the line Finish:).

Pass 1 — building the symbol table

1. ORG directive: tells the assembler to load the first instruction at that address (e.g. ORG #0100 ⇒ first instruction at address 100). With no ORG the program starts at relative address 0.
2. Read line by line, counting the address (+1 per line).
3. A label on a line (e.g. StartProg:) → enter label + that line’s address.
4. A label used in an operand before its line has been reached → enter it with UNKNOWN (address not known yet).
5. When the assembler later reaches the line that defines it → cross out UNKNOWN and write the real address.

Pass 2 — building the executable

• Uses the symbol table to replace every symbolic operand with its address.
• Uses the opcode table (the second table needed) to replace each mnemonic with its binary op code → machine code.

7. Question (b) — Trace Table with LDV (Relative Addressing)

Dry run — one round per column

Whole output: C 3 X

Completed trace table:

8. Question (c) — Symbol Table & Explanation

Board’s part-way table (first 11 entries) — what each circle means

9. Question (e) — Relative-Address Symbol Table

Order of entries (circles)

1. ① ② StartProg is the first label ⇒ relative address 0.
2. ③ ④ LDV #Offset: Offset not seen yet ⇒ enter with UNKNOWN.
3. ⑤ ⑥ CMP Value: Value not seen yet ⇒ UNKNOWN.
4. ⑦ ⑧ JPE EndProg: EndProg not seen yet ⇒ UNKNOWN (4th row of the table).
5. ⑨ reaches line EndProg: END = line 8 ⇒ UNKNOWN → 8.
6. ⑩ reaches Offset: (first data line) = line 9 ⇒ UNKNOWN → 9.
7. ⑪ reaches Value: (last data line, 6 data items 9–14) ⇒ UNKNOWN → 14.