CSE 655: Introduction to the Principles of Programming Languages

Course Summary: Studying programming language constructs, and design and implementation issues for different language families; grammars and parse trees; interpretation versus compilation; data types, binding and scope rules; language constructs for control and data abstraction.
Prerequisites: CSE 560 and CSE 625.
Credits: 4
Quarter: Au '09
Place and Time:
- DL 713; M,W,F at 11:30 am (Section: 0020-LEC(5710))
  Final exam: Thurs., Dec. 10, at 11:30;
Instructor: Neelam Soundarajan
Office: DL 579; tel: 292-1444.
Office hours: M, W, F: 12:30 pm - 1:20 pm;
Other hours by appointment.
E-mail (to neelam AT cse.ohio-state.edu) questions strongly encouraged.
If you do send me e-mail, please send only plain text or .pdf, not .doc files .html, etc.
Grader:
- Name: Dacong Yan
- Off. hrs: M, W, F: 10:30-11:20 am; and by appointment.
- Office: DL 674
- email: yan AT cse
- Phone: 292-8234
Links:
Course Notes:
Assignments:
- Assignment 1 (Due: Oct. 9, '09) (pdf)
  Note: If you read the text or other references on BNF, you might see that they use what is often called "extended BNF". What we are using in class is "pure BNF". We want to stay with pure BNF because that makes writing parsers (based on a given BNF) easier.
  So what is the difference between the two? Consider the grammar for unsigned numbers we saw in class (I am using bold face for non-terminals because .html causes trouble with angle brackets):
      no ::= d | d no
      d ::= 0 | 1 | 2 | ... | 9
  That is pure BNF. In extended BNF, you could have the same grammar since any pure BNF grammar is also, by definition, an extended BNF grammar. But you can also write it as:
      no ::= d{d}
  The notation {...} in extended BNF means, "zero or more occurrences of whatever is inside the braces". Here all we have inside the braces is d. So it means zero or more digits. (Note that the production for d is the same as before).
  Why write it like this? Because it eliminates the recursion (on the right side of the production we don't any longer have no). You cannot eliminate all recursion in BNF productions using extended BNF in this manner but you can get rid of a lot. Some people don't like recursion so they think such extended BNF productions are simpler. But, as I said, we won't use them because they cause problems for the kind of parser we will use.
  So don't use extended BNF in your answers.
  By the way, extended BNF also provides some additional facilities such as specifying that a given item in a particular alternative of a production is "optional". The key question is, what does an alternative on the right side of a production look like? In pure BNF, the answer is simple: an alternative in a production is a (finite) string of terminal and non-terminal symbols. In extended BNF, you have additional possibilities such as "{ ... }". None of this increases the power of the formalism - if can write a grammar in extended BNF, you can mechanically generate an equivalent grammar in pure BNF. Email me if you have any questions.
- Assignment 2 (Due: Oct. 23, '09) (pdf)
- Assignment 3 (Due: Nov. 20, '09) (pdf)
Labs:
- Interpreter project, part 1 (Tokenizer) (Due: Nov. 11, '09) (pdf)
- Interpreter project, part 2 (the rest) (Due: Nov. 18, '09) (pdf)
Useful info/tips:
- Tutorial on GDB (a must read if you don't regularly use a debugger).
- Info on GDB (extracted from Unix Programming Tools) (Original at: Stanford)
- Compiling with g++: Here is the syntax for compiling a bunch of C++ files using g++:
```
   g++ -o main main.cpp file1.cpp file2.cpp ...
```
- Java tip (based on a student post to course newsgroup in an earlier offering of the course):
```
To get code running with java 1.6 on stdsun you'll need to use the
'subscribe' command to subscribe to JDK6, but then you'll receive some
heap size errors. I emailed the cse help desk about it and got the
following response:

	"You can adjust your heap size by invoking the settings
	parameters. They are:

	-Xms512k (memory heap start size 512 kilobytes)
	-Xmx1024m (memory heap max size 1024 megabytes)

	Both should be invoked at the same time and you may test
	different heap sizes. Our recommendation is 16m to start and
	256m to max."

What worked for me (I couldn't figure out a way to set this
permanently), was to use java calls as "java -Xms16m -Xmx256m somefile".
```
- Here is another Java tip (for Eclipse users)
- The authoritative book on Scheme is one written Abelson and Sussman who designed the language. The title of the book is "Structure and Interpretation of Computer Programs". It is available for free on-line. But note that this book contains far more material than is included in this course.
- Some Lisp applications
- Some Lisp functions
- Here is an interesting paper on programming languages that students ought to be proficient in and related topics.
- Comparing languages (mainly with respect to speed but also expressiveness;
  Another comparison (with respect to speed, memory use etc.).
Announcements:
- The second midterm is on Monday, Nov. 23. I will post a midterm review after class on Friday.
  Here is a sample midterm; here are possible answers to the sample midterm questions.
  The sample midterm has a problem: all the questions are related to Scheme/functional programming. In the actual midterm, you can expect some of the questions to be based on our discussion of the "object-oriented" approach to the Core interpreter that we had after the first midterm.
  Note also that, as in the first midterm, you may use a single cheat-sheet (one 8" X 11" sheet, possibly doublesided, containing whatever notes you want).
- The third homework is supposed to be due in class on Friday, Nov. 20. But you can give it to me by 5 pm on Friday without any late penalty. If you want to do that (instead of giving it to me in class), please give it to me in my office (DL 579) or slip it under my office door if I am not in when you stop by, before 5 pm on Friday.
- No class on Wed., Nov. 11 (Veterans Day). The Tokenizer lab is still due at 11:59 pm on Nov. 11. The rest of the interpreter is due on Nov. 18.
- First midterm on Wednesday, Oct. 28. Topics will be everything we discuss in class by that date.
  You may use a single cheat-sheet (one 8" X 11" sheet, possibly doublesided, containing whatever notes you want).
  Here is sample midterm (but don't expect the actual midterm to be exactly like this); here are possible answers to the sample midterm questions.
- No class on Monday, October 5.
Other:
- Course newsgroup: Make it a habit to read the course newsgroup cse.course.cse655 regularly. Important announcements regarding labs, exams, etc. will often appear there. If you are having trouble reading the newsgroup from home etc., you may want to consult this document (url: http://www.cse.ohio-state.edu/~neelam/courses/655/newsgroupSettings.html).
  Here are some newsgroup postings.
- Please check your CSE e-mail regularly. And if you prefer to read mail on some on other account, use the "forward" mechanism to have all your CSE mail forwarded to that other account. Here is how you can do this (information from the CSE web site):
```
If you would rather forward your e-mail to your Ohio State e-mail 
(lastname.##@osu.edu), simply create a file called .forward in your unix 
home directory. In this .forward file, type the e-mail address you wish to 
have your CSE e-mail forwarded to and save the file. Your e-mail will then 
be forwarded to the address you specified.
```
  Again: Please read your CSE e-mail regularly.
  If the grader has problems with your assignments or running your labs etc., he/she will send mail to your CSE address and it is your responsibility to respond promptly.