University of Virginia, Department of Computer Science
CS655: Programming Languages, Spring 2001

Manifest: Tuesday 6 March 2001
Assignments
Now Problem Set 3

No class on Thursday.

Project Schedule

Your grade on the project doesn't depend on being successful as a research project. For example, if your project involved extending a language with a new feature you might propose to evaluate it by taking some existing programs and rewriting them to use the new feature and comparing the size, performance and readability of the original and rewritten programs. If your experiment concludes that the extension you designed is not a good idea, this does not necessarily reflect poorly on your project grade. In fact, it could be a great project if you can draw conclusions about why it was unsuccessful.

The best projects should lead to conference papers and contribute to your thesis. All project proposals should describe work that could lead to a conference paper if successful.

Readings
Readings for Tuesday (March 20) class: As you read the Abstraction Mechanisms in CLU paper, attempt to answer the following questions (you don't need to turn anything in, but do think about the answers):

p. 230: cvt is an unnecessary but useful syntax for making programs more compact. CLU automatically defines:

up = proc (rep) returns (wordbag)
down = proc (wordbag) returns (rep)
Rewrite the implementations of wordbag$create and wordbag$insert without using cvt.

p. 231, footnote 3: Explains why it is necessary for wordbag$insert to return a value. Could a different representation be used to make this unnecessary?

p. 231, Figure 4: Unlike Pascal and Algol68, CLU is able to define a recursive data representation without the need for explicit references. What is it about CLU's semantics that allows this? (Hint is on next page.)

p. 232: Is it possible to implement a mutable data type using an immutable representation? Is it possible to implement an immutable data type using a mutable representation?

p. 232: "These restrictions, plus the restriction that only the code in a cluster may use cvt to convert between the abstract and representation types, ensure that the behavior of an object is indeed characterized completely by the operations of its type." Is this correct? Are any other requirements on data type implementations necessary to guarantee this?

p. 233: If you already have experience with an object-oriented language that supports subtyping, consider the differences between subtypes (or Java interfaces) and where clauses. (We'll read a paper on this later.)

p. 234-235: Imagine writing the same program using Java or C++.

p. 236-237: Should data abstraction fundamentally make a program execution slower or faster?

The Abstraction Mechanisms in CLU paper was one of five papers from the Conference on Language Design for Reliable Software held in March 1977 selected for publication in Communications of the ACM. In his introduction to the special issue, Jim Horning writes:

Although the five papers in this special issue are based on particular languages, each raises issues that are of more general concern, and worthy of the attention of anyone concerned with reliable software. They were selected on the basis of their quality, but they also provide a good indication of the major themes of the conference as a whole. The programming community must soon come to terms with the topics that they address, including: What are the qualitative and quantitave effects of strong type-checking? How do verification considerations affect language design? What abstraction mechanisms should languages provide? How can security of high-level languages be extended to real-time applications?
How much progress have we made on any of these questions since then? Should we be asking different questions today?

Links

CLU Home Page (including a portable CLU implementation and CLU Reference Manual).

Questions
  • What does it mean for a language to have latent or manifest types? What are advantages and disadvantages of each?
  • What does it mean for a language to have static or dynamic type checking? What are advantages and disadvantages of each?
  • What does it mean for a language to have weak or strong type checking? What are the advantages and disadvantages of each?
  • Describe the type system of languages you know.
  • How to reflect typing rules using operational semantics.
  • How to create typing rules corresponding to an informal language description.
  • How to construct a proof that a language fragment is well-typed.

Algol 60 was not only an improvement on its predecessors, but also on nearly all its successors.
C. A. R. Hoare


CS 655 University of Virginia
Department of Computer Science
CS 655: Programming Languages
David Evans
evans@virginia.edu