CSI Fall 2002 Dr. William A. Maniatty Assistant Prof. Dept. of Computer Science University At Albany Programming Languages and Systems Concepts Fall 2002 Monday Wednesday 2:30-3:50 LI 99

Administrative Stuff Grading Policy Grading Scale as per Syllabus 2 Exams, 25 % each (Midterm and Final) 4 Projects, 10% each 1 Presentation 10% This is a Graduate Course, for Researchers Oriented to CS grads lacking CS undergrad Review Classical and Current Literature Explore state of the art

Goals of the Course What is a programming Language? A notation used to express programs (source code) A framework in which programs run Compilation - Program translated into another language and executed. Interpreter - Program is evaluated on the fly. Not the Operating System (Run time).

Prerequisites Mathematics Graph Theory Combinatorics Computer Science Automata Theory Computer Architecture Data Structures/Algorithms Miscellaneous Technical Writing

Who Should Take this Course? To succeed you should have: Motivation -Willing to work hard Preparation - Have good background Aptitude - Ability to learn and get results Should have a B or better average in Data Structures Discrete Math Computer Architectures

Academic Honesty Our reputation is all we have got. I want my students to do well. We Need Employer's and School's Trust Cheaters subject to failure and sanctions. So Protect our Reputation Try projects early If stuck see TA or the Prof. Set an example and encourage integrity Don't panic, some assignments are hard To test limits of the best students.

Design Principles Technical Factors Ease of use Problem type (Form Follows Function) Expressiveness (How general is it?) Ease of writing good interpreter/compiler Performance Flexibility/Evolution (established languages) Non Technical Factors Inertia Large Supporters/Visibility (new languages) Compiler/Interpreter Availability Personal Preference/Evangelism

Programming Language Taxonomy  Imperative (Focus on Control Flow)  Procedural (von Neuman)  Object Oriented  Declaritive  Functional  Data Flow  Logic, Constraint Based

Imperative Languages  Focus on Control Flow (Instructions)  Procedural (von Neumann)  Describes Actions on data  Assembly, Fortran, Basic, Pascal, C, Bourne Shell  Object Oriented  Language Support for Grouping Data and Operations Together (Encapsulation)  Simula 67, Small Talk, C++ (Hybrid), Eiffel, Java

Declarative Languages  Declarative = Data Driven  Functional - Based on Churches Lamda Calculus: Lisp, ML, Haskell  Data Flow -Pipelined data operations  Logical, Constraint Based - Give rules and initial condition, derive path to goal.  Prolog and Spread Sheets (Visicalc/Lotus/Excel)  Relational -Database Query - SQL

Why Should I study Programming Languages?  To allow Informed Design Decisions  Gives insight when debugging  Permits effective use of compilers/linkers interpreters and language oriented tools.  Helps to understand how langauge features work.  Learn features, emulate missing features.

Binding Time  Binding assigns values to language objects  Instruction Addresses  Data Values  Data Addresses  Binding can be  Early - Performance improved  Late - Increased Flexibility

Compilers Vs. Interpreters  Is Translation Separate from Execution?  Yes -Compiler  No - Interpreter  Combined Approach Often Used (Java)

Why Interpret?  Flexibility (provided by late binding)  Run Time Environment Support  Scripting (Perl, Shells, Python,TCL)  Dynamic Environments (Basic, APL, LISP)  Virtual Machines (JVM, Emulators, CPUs).

Why Compile?  Fundamental Engineering Principles  Correctness -Early static error checking  Cost -Can reduce cost of code distribution  Performance - Make the common case fast  Compile Once (Cost), Run Many Times (Benefit)

Multi-Pass Compilers  How to handle complexity?  Libraries (keep language simple, e.g. Java)  Layering (Focus on one problem at a time)  Results in Multiple (pipelined) phases

Intermediate Code  Some Optimizations easier at that level  Portability Easier (Pascal)  Intermediate Code Can be Interpreted

Target Languages  Many compilers emit assembly code  Can be highly optimized  Others emit higher level langauges  Exploits existing optimizers  Increases Portability, reduces complexity

Phases of Compilation  Layering induces phases of compilation

An Example  Consider the Pascal Program

Syntax Analysis (Front End)  Scanning identifies terminals (tokens)  Parsing identifies nonterminals

Semantic Analysis  Semantic Analysis is back end  Uses Abstract Syntax Tree (AST)

Optimization  Goal: Reduce Resource Consumption  Memory (data and/or instructions)  Run Time  Golden Rule: Never break working code.  Sad Truth: Most programs are broken.  No guarantees about broken programs

Summary  We will focus on imperative languages  They are by far more common  But we will look at Declaritive approaches too  Want to understand design and implementation  Explore common techniques  Often with imperative language application