CPSC 539K - What Do Programs Mean?

Winter Session 2008-2009 Term 1
Instructor: Gregor Kiczales
Meeting time: MW 9:30-11:00 ICCS 238
FIRST CLASS MEETING: Monday, September 8th, 9:30, ICCS 238.

Announcements, readings, handouts etc.

• NOTE: First paper for next couple of weeks has changed. See below for updated list.
  
  For the next several weeks we will read and discuss material on aspect-oriented programming. Our main goals in reading this material will be to understand how the various mechanisms the papers propose provide flexibility in terms of perspective from which the program can be designed/viewed/edited/compiled etc.; and to understand how that flexibility relates to the needs identified by Kent, Suchman and Smith. We want to know what aspect-oriented programming and its family of join point models do and don't do to address the problems we have been looking at.
  
  Class will not meet again until October 27th. (the 13th is Thanksgiving, the 15th I have an outside meeting that I can't miss or change, the 20th and 23rd I will be at the OOPSLA conference.)
  
  Between now and then, please read the four papers listed below. As you read them, consider the above questions, and in particular, consider issues of flexibility in individuation criteria, equivalence criteria and sharing as discussed by Kent and Smith; flexibility in terms of contextual dependence, situation particulars and incomplete description as discussed by Kent, Suchman and Smith; and flexibility in terms of coping with the kinds of ontological problems identified by Kent and Smith.
  
  Please arrive in class on the 27th having carefully read the papers and considered the above issues in general for all the papers. In addition, arrive with a specific example of how one of the mechanisms does provide a previously identified flexibility as well as an example of how one of the mechanisms fails to provide a previously identified flexibility. Hand a short description of each of these in before class as your elaborated question.
  
  The papers are:
  
  
  • N degrees of separation: multi-dimensional separation of concerns
    Proceedings of the 21st international conference on Software engineering
    Peri Tarr, Harold Ossher, William Harrison, Stanley M. Sutton, Jr.
    From ACM Digital Library.
  • Aspect-oriented programming
    Proceedings of the European Conference on Object-Oriented Programming (ECOOP) 1997.
    Gregor Kiczales, John Lamping, Anurag Mendhekar, Chris Maeda, Cristina Videira Lopes, Jean-Marc Loingtier, John Irwin.
    From my personal web page.
  • An overview of AspectJ
    Proceedings of the European Conference on Object-Oriented Programming (ECOOP) 2001.
    Gregor Kiczales, Erik Hilsdale, Jim Hugunin, Mik Kersten, Jeffrey Palm and William G. Griswold.
    From my personal web page.
  • Minimizing dependency on class structures with adaptive programs Proceedings of ISOTAS-93.
    Karl J. Lieberherr and Cun Xiao
    From Springer.

---

• On October 8th, Richard P. Gabriel will be our guest lecturer, the topic will be Christopher Alexander's new book The Nature of Order. Class will meet from 5-8 pm in DMP 101. There will be no class October 6th. Dinner will start at 4:15 in room 206.
• For September 29 and October 1, we will read the Introduction and Chapter 1 of On the Origin of Objects. I suggest the following three part strategy for reading the introduction.
  • First spend no more than 15 minutes skimming it.
  • Then do a little web-work and get yourself working definitions of the following terms: metaphysics, realism, constructivism, pluralism, cognitivism, ontology and epistemology.
  • Then go through the material again carefully, identifying the individual claims the author is making and trying to build up a picture of the intellectual claims he is making.
  I also suggest trying to get a first pass of Chapter 1 done over the weekend rather than waiting until after class Monday.
• For Monday September 22nd read Chapter 5 of Plans and Situated Actions; for Wednesday read Chapter 6.
• For September 17th, read Chapters 4 and 5 of Data and Reality.
• For September 15th, read Chapters 2 and 3 of Data and Reality.
• For September 10th, read the Preface and Chapter 1 of Data and Reality.
• Effectiveness w/o Formality slides: for viewing, for printing.
• Course intro, process etc. slides.

It is hard to overstate the success of the software revolution. Software systems are everywhere: cars, phones, banks, taxis, hospitals, courier services, copiers, radios and so on... and on and on... and on.

And yet users as well as producers (managers, strategists, designers, programmers etc.) of software systems all complain about inflexibility, brittleness and other difficulties in getting the systems to do what people want. Years of research on techniques for improving software tailorability and flexibility has produced only modest results.

Why are software systems so inflexible? Why is it so hard to change them to work in different ways or to integrate into different contexts? Why is it so hard to understand what they do, why and how that might change in a different context? Why is it so hard to even get a handle on the problem?

In this course we will explore whether the problem might be more fundamental than has previously been acknowledged. We will do so by trying to go to the root of one of the most basic questions in the field: what do programs mean?

The traditional answer is that programs are crisp and well-defined instructions for a computer to follow to produce a desired process and result. Most work in programming language foundations is directed at this mechanistic sense of program meaning.

A different kind of answer is that programs embody an understanding of some part of the world (real and/or imagined); and are intended to express that understanding in different ways to their authors, readers and users.

This expressive sense of program meaning is at least partly odds with the mechanistic sense, because people do not see the world in crisp and well-defined ways. Instead our understanding of the world seems to be imprecise, transient, negotiated and many things other than crisp. The underlying world itself may not even be crisp and well-defined in the first place.

The course explores two broad questions.

The first is to ask what kind of theoretical underpinnings could help us understand and better work with the expressive sense of program meaning. What do we mean when we say people see the world differently? Or that the world is not crisp and well-defined? What form(s) might a foundational theory of program expressiveness take? What other disciplines could we appeal to for inspiration? What existing questions might such a theory help us answer?

The second question seeks to resolve existing mechanistic foundations with possible new expressiveness foundations. In particular, we will ask whether an improved understanding of the ways in which programs are (and need to be) expressive also suggests new mechanisms.

The course will include readings excerpted from several books and technical papers, class lecture and discussion, and work with a simple tool for prototyping alternative programming language mechanisms. Each student will complete a project due at the end of the semester.

The course is not intended to be overly difficult. There will not be huge amounts of reading nor will there be a massive implementation project. But please bring an open mind to (re)consider questions of what programs are and mean.

• Selections from William Kent, Data and Reality.
• Selections from Lucy Suchman, Human-Machine Reconfigurations: Plans and Situated Actions.
• Selections from Christopher Alexander, The Timeless Way of Building.
• Selections from Brian Cantwell Smith, On the Origin of Objects.

In addition, a modest number of technical papers will be made available on-line as the course progresses.