2018-10-29

programming paradigms

brief overview

what is a paradigm

a paradigm is a typical example or pattern of something; a pattern or model. the programming paradigms we are going to look at are:

imperative

  • procedural
  • object-oriented

declarative

  • functional
  • logic

note about languages

languages usually support a mix and possibly all programming paradigms. some languages are designed to favor one paradigm. for example java with object-oriented programming. some ways in which languages make it more difficult to use alternative paradigms are:

requirements

for example requiring to create files with classes to be able to create functions

limitations

unsupported features, for example no support for first-class functions domain- and problem-specific built-ins

short descriptions

imperative

series of commands that modify a programs state

procedural

imperative commands structured into subroutines

object-oriented

structure a program into objects with behaviour and data. or: both subroutines and data stored in records

declarative

describe "what" and not exactly "how"

functional

treat computation as the evaluation of mathematical functions

logic

express facts and rules and ask the system

additional descriptions

imperative

  • the order of expressions is usually very relevant. this can be problematic for parallelism and the understanding of code
  • relatively few abstractions away from the machine (von neumann achitecture), this makes manual low-level performance optimisations easier to make

procedural

avoids having to deal with goto by using subroutines

object-oriented

avoids having to deal with subroutines that modify global state by encapsulating data in classes and objects

functional

  • ideally no side-effects (does not mutate anything outside of the function)
  • order of computation is less relevant as there is less hidden shared state. this is beneficial for parallelisation because it makes it simpler to evaluate function calls independently on separate processors
  • referential transparency (same input leads to same result, function and specific arguments equal one specific result, function calls an variables can be substituted with results)
  • more predictable (formal verification, automatic optimisations)
  • state changes are always explicit via return values and function arguments

logic

  • logic and control
  • leaves much of the "how", the choice of necessary computations, to the system

code examples

the factorial function implemented in different styles. each example shows definition then application. the factorial function calculates the product of all positive integers less than or equal to n

example

factorial(5) = 5 * 4 * 3 * 2 = 120

procedural (coffeescript)

factorial = (n) ->
  result = 1
  while n >= 1
    result = result * n
    n = n - 1
  result
factorial 5

object-oriented

class Factorial
  result: null
  calculate: (n) ->
    @result = 1
    while n >= 1
      @result = @result * n
      n = n - 1
f = new Factorial
f.calculate 5
f.result

functional

factorial = (n) ->
  if n <= 1 then 1 else n * factorial(n - 1)
factorial 5

alternative

factorial = (n) ->
  [1..n].reduce (result, n) -> result * n

logic (prolog)

factorial(0,1).
factorial(N,F) :-
  N > 0, N1 is N - 1, factorial(N1,F1), F is N * F1.
?- factorial(5, X).