Systematic Development of Programming Languages

1. Systematic Development of Programming Languages L1Intro

2. Orthogonal Parameters • Conceptual view (Model of Computation) • imperative, functional, relational,... • Level of abstraction (Model of Implementation) • problem domain • ... • machine • Computational Task : I/O relation to be implemented L1Intro

3. Ij 1. Customized Digital Computer Task j Oj Task k Ik Ok Rig different circuit for each task L1Intro

4. 2. Stored Program Computing • von Neumann showed the existence of a Universal Machine (hardware) that can be customized using control inputs to carry out different tasks. • Software is the encoding of the task to control this machine. encoding(Tj) Ij Oj L1Intro

5. Imperative Languages • Model of Computation • ALU + Memory + Input + Output (von Neumann architecture) • Levels of Abstraction (“Human Interface”) • Machine Language • Binary Representation of the task • Assembly Language • Symbolic Representation of the task L1Intro

6. Assembly Language ADDI R4 R2 21 ADDI R4,R2,21 10101100100000100000000000010101 • Use symbols instead of binary digits to describe fields of instructions. • Every aspect of machine visible in program: • One statement per machine instruction. • Register allocation, call stack, etc. must be managed explicitly. • No structure: everything looks the same. L1Intro

7. Pros and Cons of Assembly Language • Avoids Absolute Addressing • relocatable, reusable/shareable • Uses Symbolic Names • readable • Low-level programming wastes effort in coding a solution rather than solving a problem. • Difficult to build and maintain large programs. L1Intro

8. High-level Language • Provides notation to describe problem solving strategies rather than organize data and instructions at machine-level. • Improves programmer productivity by supporting features to abstract/reuse code, and to improve reliability/robustness of programs. • Requires a compiler. L1Intro

9. Problem Domain (stacks, tables) Java/C#/Scala/Python/Scheme/Clojure (Class Hierarchies) Levels of Abstraction Ada (ADTs) (arrays) Pascal C++ C Assembly Language Machine (char, int) L1Intro

10. Evolution of Programming Languages • FORTRAN ( FORmula TRANslator) Goals : Scientific Computations Efficiency of execution Compile-time storage determination Features : Symbolic Expressions Subprograms Absence of Recursion • COBOL Goal: Business Application Features : Record/Structure; File Handling L1Intro

11. Evolution of Programming Languages • ALGOL - 60 (ALGOrithmic Language) Goals : Communicating Algorithms Features : Block Structure (Top-down design) Recursion (Problem-solving strategy) BNF - Specification • LISP (LISt Processing) Goals : Manipulating symbolic information Features : List Primitives Interpreters / Environment L1Intro

12. Problems • Not rugged wrt typographical errors Do 10 I = 1.20 I5 = I5 + 1 vs vs Do 10 I = 1,20 I5 = IK + 1 • Remedy: Use spaces to delimit tokens, Declare before use • Unintended Coercion I > J and false • Remedy: Type checking L1Intro

13. Evolution of Programming Languages • Pascal Goal : Structured Programming, Type checking, Compiler writing. Features : • Rich set of data types for efficient algorithm design • E.g., Records, sets, ... • Variety of “readable” single-entry single-exit control structures • E.g., for-loop, while-loop,... • Efficient Implementation • Recursive descent parsing L1Intro

14. Programming in the Large • Programs no longer monolithic. Developed by a team of programmers. • Code sharing and reuse very important. • Correctness, reliability, and robustness essential. • Data Abstraction / Encapsulation / Strong Typing • Ada, CLU, Modula etc. L1Intro

15. Other Languages • Functional • Common LISP, Scheme, Clojure • ML, Haskell • Logic • Prolog • Object-oriented • Smalltalk, SIMULA, Modula-3, Oberon • C++, Eiffel, Ada-95, Java, C# • Hybrid • Python, Ruby, Scala • Application specific languages and tools L1Intro

16. Designed for gluing applications : flexibility Interpreted Dynamic typing and variable creation Data and code integrated : meta-programming supported Examples: PERL, Tcl, Python, Ruby, PHP, Scheme, Visual Basic, etc. Designed for building applications : efficiency Compiled Static typing and variable declaration Data and code separated : cannot create/run code on the fly Examples: PL/1, Ada, Java, C, C++, C#, Scala, etc. Scripting vs Systems Programming Languages L1Intro

17. (cont’d) • Does application implement complex algorithms and data structures? • Does application process large data sets (>10,000 items)? • Are application functions well-defined, fixed? If yes, consider a system programming language. • Is the main task to connect components, legacy apps? • Does the application manipulate a variety of things? • Does the application have a GUI? • Are the application's functions evolving rapidly? • Must the application be extensible? • Does the application do a lot of string manipulation? If yes, consider a scripting language. L1Intro

18. Jython (for convenient access to Java APIs) I:\tkprasad\cs7100>jython … >>> import javax.swing as swing >>> win = swing.JFrame("Welcome to Jython") >>> win.size = (200, 200) >>> win.show() >>> ^Z L1Intro

19. map = new HashMap(); map.put("one",new Integer(1)); map.put("two",new Integer(2)); map.put("three",new Integer(3)); System.out.println(map.get("one")); list = new LinkedList(); list.add(new Integer(1)); list.add(new Integer(2)); list.add(new Integer(3)); map = {"one":1,"two":2,"three":3} print map ["one"] list = [1, 2, 3] Java vs Jython L1Intro

20. for i in list: (* iterator *) newList = [function(i) for i in oldList] (* list comprehension *) for (Iterator i; i.hasNext();) { i.next(); } List newList = ArrayList() for (Iterator i; i.hasNext();) { Object obj = i.next(); newList.add(function(obj)) } (cont’d) L1Intro

21. Functional Programming in Jython apply(lambda x,y : x*y, (10, 20)) # 200 map(lambda x,y: x + y, [[1,2],[“a”]], [[“3”],[“b”]]) # [[1, 2, ‘3’], [‘a’, ‘b’]] reduce(lambda x,y: x + y, [1,2,3], 100) # 106 filter(lambda x: x > 0, range(10,-5,-3)) # [10, 7 , 4, 1] L1Intro

22. Meta-programming in Jython • Dynamic code evaluation print eval (“[1,3] + range(6,10,3)”) # [1, ,3, 6, 9] x = 2 + 3j exec “x = 5, x + x” #(5, (4+6j) L1Intro

23. Java functionality through Jython import java.lang as lang import javax.swing as swing import java.awt as awt names = ["Groucho", "Chico", "Harpo"] quotes = {"Groucho": "Say the secret word", "Chico": "Viaduct?", "Harpo": "HONK!"} def buttonPressed(event): field.text = quotes[event.source.text] def exit(event): lang.System.exit(0) def createButton(name): return swing.JButton(name, preferredSize=(100,20), actionPerformed=buttonPressed) L1Intro

24. win = swing.JFrame("Welcome to Jython", size=(200, 200),windowClosing=exit) win.contentPane.layout = awt.FlowLayout( ) field = swing.JTextField(preferredSize=(200,20)) win.contentPane.add(field) buttons = [createButton(each) for each in names] for eachButton in buttons: win.contentPane.add(eachButton) win.pack( ) win.show( ) L1Intro

25. Current Trend • Multiparadigm languages • Functional constructs for programming in the small • Focus on conciseness and correctness • Object-Oriented constructs for programming in the large • Focus on programmer productivity, robustness and code evolution • Example languages • Older: Python, Ruby • Recent: Scala, F# L1Intro

26. Scheme (dialect of LISP) • Recursive definitions • Symbolic computation : List Processing • Higher-order functions • Dynamic type checking • Functional + Imperative features • Automatic storage management • Provides a uniform executable platform for studying, specifying, and comparing languages. L1Intro

27. Standard ML and Scala • Strongly typed language • static type inference • SML supports polymorphic types • Supports Abstract Data Types and Modules • Higher-order functions • Pattern matching • cf. Prolog, list-processing in Scheme L1Intro

28. Java vs Scala //Java - what we're used to seeing public String buildEpochKey(String... keys) { StringBuilder s = new StringBuilder("elem") for(String key:keys) { if(key != null) { s.append(".") s.append(key) } } return s.toString(). toLowerCase() } L1Intro

29. Java vs Scala //Scala def buildEpochKey(keys: String*): String = { ("elem" +: keys) . mkString(".") . toLowerCase } println(buildEpochKey("aBcDeF")) L1Intro