Software
- Harper's Regular Expression Matcher;
Computers and Software
In this course, we will be using several software systems to
develop a compiler for a variant of python. We will use the
programming language ML (an acronym for "MetaLanguage") as our
implementation language; we'll use the Emacs text editor as our
development system and the subversion revision control system to
manage the various files. Our compiler will generate native code
for the MIPS architecture.
The software required for this course will be installed on the
cslab system so you can do your work there. If you have your
own computer, you will probably want to install all of the
tools on your own computer --- they are all freely available on
the internet.
The Implementation Language: ML
ML is a statically type, value-oriented programming language
that is especially well-suited for compiler writing. (ML is
usually referred to as a "functional" programming language but
the adjective "functional" is a little misleading and doesn't
really capture the basic idea behind this class of programming
languages.)
There are two major dialects of ML:
Caml and
Standard ML. We will be using the
Objective
Caml variant of the Caml programming language. Don't be
fooled by the "Objective" in "Objective Caml" --- we're using
Objective Caml for its advanced module system and software tool
suite. We won't be using the object system. An implementation of
Objective Caml can be found on a
server, yes, in
France.
- Objective Caml has an excellent
User's Guide
which should prove to be helpful in coming up to speed on
programming in Objective Caml. It certainly deserves a bookmark.
- There are a number of
good tutorials on
Objective Caml.
- Jason Hickey has written a very good draft of a textbook
on Objective Caml:
An Introduction to Objective Caml.
The Editor: GNU Emacs
Emacs is an
extensible and customizable text
editor that has been kicking around the Unix community for about
30 years and has something of a cult following. An implementation
of Emacs is available on the lab computers and is part of the
standard linux and OS X distributions.
If you would like to use your own computer for this course but
you do not have Emacs on your system, you can find numerous
implementations for various platforms on the web. I recommend
installing
GNU Emacs.
See me or Phil Temples if you have trouble with the installation.
If you are new to Emacs, run it from the command line and then
take the built-in tutorial by typing "ESC x help-with-tutorial".
When the Emacs editor boots up, it looks for a file called
.emacs in your home directory. This file contains initialization
code (written in LISP). The LISP code provides hooks into Emacs
which can be used to implement specialized "modes" for editting
different languages. We will be using a customized mode for
Caml called tuareg.
The code implementing tuareg-mode can be
downloaded here.
Follow the instructions in the README file to install
it. See me if you need help with this step.
The Revision Control System: svn
Subversion is an
open source revision control
system which allows a group to establish a repository (file
structure) on one system and synchronizes concurrent access by
users from remote systems. The repository contains a master
copy and each user "checks out" a local working copy of the
master.
We will use svn for managing our files, for distributing support
code and interfaces and for submitting work for grading. We
will provide subversion repositories for each of you on the
cslab system.
If you would like to use your own computer and it doesn't have
subversion, you can freely
download an implementation.
The svn Quick Reference Card
may be worth a bookmark.
The Target Language: MIPS
Our compilers will target the MIPS architecture by emitting code
in (a subset of) the MIPS assembly language. Most of you
are probably familiar with this architecture from the Computer
Organization course. The .asm files can be assembled and the
resulting binary can be run on the
MARS
MIPS Simulator. Like all of the others, this system is freely
available.
In addition to the MARS implementation, we will provide a
MIPS virtual machine in Caml so you will be able to execute
the code directly in Caml without needing to assemble and
run it using MARS.
Getting Help
Developing a compiler is a major undertaking involving thousands
of lines of code. It is especially daunting when using a new
programming language for the implementation. I expect to spend
a good deal of time with students helping track down bugs and
working on design issues. I suggest working in groups in the
CS office suite so you can stop in with questions.
|
