Do not send mail other than program submissions to cs302acc, as it is not regularly read.
Instructor:
Andrew Tolmach
PCAT 128A
725-5492
email: apt@cs.pdx.edu
Office Hours: MW1-2 & by appt.
Course home page: http://www.cs.pdx.edu/~apt/cs302
CS301/302 studies, in parallel, the design and implementation of programming languages and the design and implementation of compilers. The course is centered around a substantial programming project: implementing a complete compiler for a realistic language. CS302 concentrates on code generation and runtime organization; we will assume the existence of a compiler front-end, and build a back-end for it. Completed compiler projects will produce machine code that can be run directly on the target hardware. A detailed list of lecture topics may be found on the accompanying schedule sheet.
CS301 with a grade of C or better, preferably taken Winter 99 with this instructor. CS301 and 302 form a connected pair of courses; having completed CS301 in the distant past with another instructor will not necessarily have prepared you for this version of CS302.
Substantial experience with the C or C++ programming language and the Unix operating system is essential. Some experience with machine-language programming is extremely desirable.
The required textbook is Aho, Sethi, and Ullman, ``Compilers: Principles, Techniques, and Tools,'' Addison-Wesley, 1986. Additional readings from other sources will be handed out in class from time to time. A good auxiliary reference text is R. P. Paul, ``SPARC Architecture, Assembly Language Programming, and C,'' Prentice Hall, 1994.
Lecture notes will be available electronically in postscript format (and usually also in html) via the course home page.
Over the course of CS302 you will complete the PCAT compiler you began in CS301. You will also build an interpreter for the language. A working version of the front end will be available for you to use in place of your own, if you prefer.
You are strongly encouraged to work in teams of two on the projects. Team members submit a single version of each project assignment, and receive the same grade. The first programming assignment of the term is not part of the project and must be completed individually.
Your compiler must be written in ANSI C or C++. Supporting code handed out by the instructor will be in C; if you use C++, you will need to work at least partly in the C subset of the language. C is described in B. W. Kernighan and D. M. Ritchie, The C Programming Language, 2nd edition, Prentice-Hall, 1988, and in numerous other reference books. C++ described in M.A. Ellis and B. Stroustrup, The Annotated C++ Reference Manual, Addison-Wesley, 1990, and in numerous other reference books. Also, you will continue to need the PCAT language definition handed out in CS301.
There will be one mid-term and a final exam. Both are closed-book. Exams will cover topics from lectures and readings, emphasizing material that is not directly relevant to the programming project. Not all the material is covered in the readings, so lecture attendance is important. Exams are scheduled in advance; unless prior arrangements are made, a grade of zero will be recorded for missed exams.
Homework problems may be assigned from time to time as an aid to help you study for the exams; they will not be collected or graded.
Approximately 1/2 on programming assignments and 1/4 on each of 2 exams (including the final).
Programming assignment grades will be determined primarily by observing program behavior on test inputs. Computer programming being what it is, this policy means that ``small'' errors in your code can have a large effect on your grade. Sample test inputs and a correctly-behaving executable will be provided for you to compare your program against. It is your responsibility to apply these tests and others of your own devising before submitting your assignment. We will apply some non-public tests to your programs as well.
For most of the assignments, you will need access to a Sparc processor, such as those found on the CS department's network. When feasible, you may develop your project solutions on any machine you like, but the final version of all assignments must work using the GNU gcc or g++ compiler on the CS department's Sparc machines. If you do not have an account on the department's Sparc network, you should obtain one immediately by following the instructions on The Form, available in class or from the CS department office.
Directory /u/cs302acc/bin will contain some useful programs; you should place this directory in your PATH.
Programs should be submitted by mailing a ``bundle'' containing the
relevant files to cs302acc@cs.pdx.edu.
Bundles can be created by the Unix shar command,
which combines the files given by its arguments into a single shell script
that, when run, recreates the files.
For example, you could submit
the files makefile, lex.c, and main.c as follows:
Do not send mail other than program submissions to cs302acc, as it is
not regularly read.
Important information will be distributed throughout the term via
a Majordomo list called cs302list. You can subscribe to this list
by sending a mail message to majordomo@cs.pdx.edu with no subject
line and a message body consisting of the single line
Please mail questions to the instructor directly (at apt) rather
than to this list; the instructor will copy mail of general interest
to the list.
All dates, including assignment deadlines, are tentative.
| Date | Asgn | ASU Reading | Topics | |
| Mar | 29 | 1 | introduction; synopsis of backend tasks | |
| 31 | Sparc | SPARC architecture | ||
| Apr | 5 | syntax and semantics revisited | ||
| 7 | 2 | interpreters; review of attribute grammars | ||
| 12 | 1* | 7.1,7.6 | environments; procedure activations and scope | |
| 14 | expressions and statements | |||
| 19 | 8.1-3 | intermediate representations; expressions and simple statements | ||
| 21 | 8.4-6 | booleans; backpatching | ||
| 26 | 8.7 | statements | ||
| 28 | 2* 3 | 7.2-4 | compiled runtime organization | |
| May | 3 | Midterm Exam (in class) | ||
| 5 | 7.5 | parameter passing mechanisms | ||
| 10 | procedures as parameters; functional programming | |||
| 12 | 9.1-3,9.11-12 | code generation; instruction selection | ||
| 17 | 4 | 9.4-6 | naive register allocation | |
| 19 | 3* | 9.7 | better register allocation | |
| 24 | 9.9,10.1-3 | optimization principles; instruction scheduling | ||
| 26 | implementing object-oriented features | |||
| 31 | No class - Memorial Day holiday | |||
| June | 2 | 4* | garbage collection | |
| 7 | 3:30pm | Final Exam | ||
Programming assignments, listed below, are distributed on the dates the assignment number appears in the schedule above and are due when the starred number appears.
| 1 | SPARC machine code exercise |
| 2 | interpreter |
| 3 | preliminary code generation |
| 4 | final code generation |
Programs are due by 3:30 p.m. on the specified due date, i.e., just before class. Late programs are not accepted except in extraordinary circumstances, and then preferably by prior arrangement. The project deadlines are there to help you, by forcing you to keep up during the term.
Unless specifically indicated otherwise, all programming assignments must represent your own, individual work. It is permissible to discuss the assignment with other students, but not to share solutions (although you can consult the tutors for help in debugging). Do not, under any circumstances, copy another person's program and submit it as your own. Writing code for use by another or using another's code in any form (even with their permission) will be considered cheating, the penalties for which are described in detail in the CS Department's booklet Undergraduate Programs 1998-1999. In particular, cheating will result in an automatic F for the assignment in question, and a letter to your departmental file.