Lecture 1 -

Logistics of the Course

- Syllabus
  - Learning objectives
  - Assessments
  - Textbook
  - Late Work
  - Collaboration

What is a Virtual Machine?

Four broad types of virtual machines

1. Hypervisor: allows (multiple) guest operating systems to run on a single machine
  - examples: VMWare, Xen, VirtualBox, Parallels
  - "cloud computing"
  - type 1 and type 2 hypervisors: bare metal and hosted on an existing OS
2. Emulator: emulates the behavior of an existing hardware architecture in software
  - examples: qemu
  - NES emulator
3. Bytecode Virtual Machine: executes a "made up" software-defined instruction set (or IR)
  - examples: JVM, CLR, p-code, o-code
4. Language Virtual Machine: executes a source code (often scripting) language
  - Python, Lua, R, JavaScript, bash

 - why a virtual machine is good (3/4)
   - less work for language designers, can evolve independently
   - portability over different hardware
   - split language frontends into separate process
   - new implementation techniques and optimizations for language features

- differences between 3 and 4
 - 4 langvm must have parser, support full language
 - langvm means the source language is a compilation target, e.g. javascript

How Virtual Machines relate to compilers
 - diagram of front, middle, back end of compiler
 - but virtual machines can also *interpret* code
 - virtual machines have a lot more services
   - garbage collection
   - debugger / profiler
   - exception handling
   - libraries that support the language
 - virtual machine is even more complicated than a compiler, has compilers inside

key terms:
 meta
   - the "representation of" or data about, something
 metadata
   - data about, or describing other data
 metaobject
   - the object which describes an object, e.g. a class object describes an instance
 intermediate representation
   - internal data structures representing code or data
   - any representation of code between source and machine code
 compile time
   - when translating source code to an intermediate or machine code form
 load time
   - when loading and linking code from disk into a running program
 run time
   - during execution of a program
 static
   - determinable without running the program
 dynamic
   - determinable or happening while running the program
 emulator / simulator
   - a software program that simulates the execution of an existing hardware architecture
 virtualization
   - managing resources in a way that makes them transparent to multiple users, e.g. multiplexing
   - implementing the semantics of a machine, bytecode, or source language on top of another machine
 "full" virtualization
   - complete implementation of semantics
 semantics
   - the complete rules for the behavior of a program or machine
 statically-typed
   - types exist in source code and apply to storage locations
 dynamically-typed
   - types exist at runtime and are attached to values
 guest language/CPU/ISA/VM
   - the language running "on top" of, or implemented by a VM
 host CPU/machine
   - the lowest level execution on which a VM is running
 implementation language
   - the language in which a virtual machine is written
 interpreter
   - executes the representation program without generating new host code
 compiler
   - translates a program from one representation to another
 JIT compiler
   - translates a program during runtime

Components of a Virtual Machine

- Code format
- Code loader
  - could be text parser, or binary loader
- Code verifier
- Intermediate representation (IR)
  - Object model
- Memory management
  - Garbage collector
- Interpreter
- JIT Compiler
- Debugger
- Profiler
- Libraries