COMPUTING SCIENCE 370 -- Programming Languages

Programming Language Paradigms

Scientific Paradigms

paradigm

"universally recognized scientific achievements that for a time provide model problems and solutions to a community of practitioners" [Thomas Kuhn, 1962]

• defined intensionally -- by their properties
• defined extensionally -- by one or more instances

  exemplar

  an example that helps to define a paradigm

Components of a Paradigm

A discipline organized around a paradigm has several components [Kuhn, 1970]:

• symbolic generalizations -- written rules or laws of the paradigm
• beliefs of the community of practitioners
• values of a group about what is most important
• exemplars, including problems and solutions

In support of the second and third points, see Kent Pitman's article More Than Just Words: Lambda, the Ultimate Political Party in which he argues that " Lisp is better defined as its community than as its various specifications."

Programming Language Paradigms

paradigms

"patterns of thought for problem solving" [Peter Wegner, 1988]

• How are they discovered?
• Why do we consider abstractions rather than individual languages?
• What are the central concepts and the exemplars for each paradigm?

Imperative Paradigms

1. Block-structured, procedural languages
   • statement-oriented, with variables holding values
   • can model the execution of a program as a series of states of variable locations
   • two kinds of statements:
     • declarations (non-executable)
       • allocate memory
       • bind symbolic names to absolute memory locations
       • initialize memory
     • imperatives (executable)
       • computation
       • control flow
       • I/O
2. Object-oriented languages
   • objects contain private state information and communicate by sending messages
   • concept of classes, with instantiations of differences
   • programming as simulation

Declarative Paradigms

"Higher-level" programming

• can express the same program with less detail -- the language does more automatically
• less procedural than a lower-level language -- we can focus on "what" instead of "how"

1. Applicative (functional) programming
   • operation-oriented: successive function evaluation
   • pure functional programming -- no side effects, no concept of locations
   • functions are first-class values
2. Logic programming
   • a series of logical assertions and queries
   • specifications can be executed directly
   • knowledge base and inference facility

Distributed Paradigms

Distributed programming may constitute a separate paradigm.

Exemplars:

• Ada (rendezvous)
• Concurrent Prolog
• Linda
• Occam