1. Lesson Objectives Aims Key Words Compiler, interpreter, assembler
To understand the differences between the three main types of translator: compiler, interpreter and assembler.
Key Words Compiler, interpreter, assembler

2. Compilation Source Code (High Level) Compiler Executable (low level)
Code runs on target system

3. Compile once – run many
Works only on target system
Self contained code (except system calls)
Protects intellectual property (unless decompiled/reverse engineered/open source)

4. To port code: Must be re-compiled for another target
Likely need to re-write (libraries, system calls etc)
Code can be cross platform but:
Executable will still only run on one system
Redundant code

5. Interpreter Interpreter targeted at specific system
Source Code
Interpreter targeted at specific system
Code runs on target system

6. Interpreters
No executable produced - it translates each line of code into machine code, line by line.
This means it can be platform independent code (the interpreter has to handle all the differences in system architecture)
Machine code produced while the interpreter is running is not saved – code is re-generated every time it is run
Used in some common languages including BASIC, Lisp, Prolog, Python, JavaScript.
Code cannot be executed without an interpreter installed on the target system.

7. Advantages Disadvantages
Source code is write once – run anywhere (on any computer with an interpreter.)
You can easily inspect the contents of variables as the program is running.
You can test code more efficiently as well as being able to test single lines of code.
Code can be run on many different types of computers and OSs, such as Macs, PC, UNIX and LINUX.
Disadvantages
You must have an interpreter running on the computer in order to be able to run the program.
An interpreter must be created for each individual computer architecture you want code to run on
As source code must be compiled each time, interpreted programs can sometimes run slowly.
You have to rely on the interpreter for machine-level optimisations rather than programming them yourself.
Developers can view the source code, which means that they could use your intellectual property to develop their own software.

8. Virtual Machines
There is a spanner in the works
There are languages that are both compiled and interpreted
Java.
Why? Why do they do it?!

9. It’s another take on the interpreted idea.
The compiler turns code into “byte code” which is an intermediary between source code and fully compiled code
Funnily enough this code is called “intermediate code”
The interpreter is then called a “virtual machine”

10. VM Method Source Code Compiler VM Code written in a given language
Turns source into intermediate code
This is not code that any real machine can run VM
Understands the “generic” intermediate code instruction set
Provides sandboxed environment
Interprets intermediate code to generate machine code

11. Why?
Compile on ANY platform
Run on ANY platform (with the VM)
Faster than pure interpreted languages

12. Assembler Assembly Language Source Code
Assembler (Op Codes)
Translator
(binary)
Executable (code runs on target)

13. An assembler is one step up from coding purely in machine code (0’s and 1’s (think MIT’s ALTAIR (one for the geeks)))
Assembly language uses mnemonics to represent cpu instruction codes
Each translates to a hexadecimal token and then a binary op code

14. Assembler matches directly to op codes
Whereas a high level language will turn single instructions in to many different op codes (and depending on compiler, different methods to solve the same problem)
Turning assembly code in to an executable is often called “assembling” rather than compiling

15. Nearly all machine code instructions are in two parts:
Opcode
Data
The op code is the instruction, the data is, er, the data!

16. Each binary opcode is usually unique to a processor
X86, X64, PPC, ARM etc may share instructions but not op codes
This explains a LOT about why code for one system won’t work on another!

17. Assemblers make life a lot easier by providing:
mnemonics
Assemblers make life a lot easier by providing:
Mnemonics
ADD
SUB
MOV
JMP
Labels
Automatic calculation of memory addresses for jumps and loops

18. And this used to be done by hand…
Compiling
Even though assembly is 1-1 instruction mapping it still takes work to compile a program
First all labels are allocated memory addresses and a symbol table created
Then the code is compiled, linked to the symbol table
And this used to be done by hand…

19. Review/Success Criteria
You should know:
The difference between assembly, interpreters and compilers
How software is turned from source code to executable