Chapter 4:

Computer Languages,
Algorithms and Program
Development

How do computers know what

we want them to do?
 Computer Languages, Algorithms
and Program Development
 In this lecture:
• What makes up a language and how do we use language to communicate
with each other and with computers?
• How did computer programming languages evolve?
• How do computers understand what we are telling them to do?
• What are the steps involved in building a program?
 Communicating with
a Computer
 Communication cycle
• One complete unit of communication.
– An idea to be sent.
– An encoder.
Speaker encodes Listener decodes
– A sender. information information
– A medium.
– A receiver.
– A decoder.
– A response.

Listener returns
feedback to speaker
 Communicating with
a Computer
 Substituting a computer for
one of the people in the
communication process.
User encodes
• Process is basically information Computer decodes
the same. information
– Response may be symbols
on the monitor.

Computer
returns results
to user
 Communicating with
a Computer
A breakdown can occur any place along the cycle...
 Between two people:  Between a person and a computer:
• The person can’t hear you. • The power was suddenly
• The phone connection is broken in interrupted.
mid-call. • An internal wire became
• One person speaks only French, disconnected.
while the other only Japanese. • A keyboard malfunctioned.

When communicating instructions to a computer, areas

of difficulty are often part of the encoding and decoding
process.
 Communicating with
a Computer
 Programming languages bridge the gap between human thought
processes and computer binary circuitry.
• Programming language: A series of specifically defined commands
designed by human programmers to give directions to digital computers.
– Commands are written as sets of instructions, called programs.
– All programming language instructions must be expressed in binary
code before the computer can perform them.
 The Role of Languages
in Communication
 Three fundamental elements of language that contribute to the
success or failure of the communication cycle:
• Semantics
• Syntax
• Participants
 The Role of Languages
in Communication
 Semantics: Refers to meaning.
 Human language:  Computer language:
• Refers to the meaning of what is • Refers to the specific command
being said. you wish the computer to perform.
• Words often pick up multiple – Input, Output, Print
meanings. – Each command has a very
• Phrases sometimes have idiomatic specific meaning.
meanings: – Computers associate one
– let sleeping dogs lie meaning with one computer
(don’t aggravate the situation command.
by “putting in your two • The nice thing about computer
cents”) languages is the semantics is
mostly the same
 The Role of Languages
in Communication
 Syntax: Refers to form, or structure.
 Human language:  Computer language:
• Refers to rules governing • Refers to rules governing exact
grammatical structure. spelling and punctuation, plus:
– Pluralization, tense, agreement – Formatting, repetition,
of subject and verb, subdivision of tasks,
pronunciation, and gender. identification of variables,
• Humans tolerate the use of definition of memory spaces.
language. • Computers do not tolerate syntax
– How many ways can you say errors.
no? Do they have the same  Computer languages tend to have
meaning? slightly different, but similar,
syntax
 The Role of Languages
in Communication
 Participants:
• Human languages are used by people to communicate with each other.
• Programming languages are used by people to communicate with machines.

 Human language:  Computer language:

• In the communication cycle, humans can
respond in more than one way. • People use programming
– Body language
languages.
– Facial expressions
• Programs must be
translated into binary code.
– Laughter
• Computers respond by
– human speech
performing the task or not!
 The Programming
Language Continuum
 In the Beginning...Early computers consisted of
special-purpose computing hardware.
• Each computer was designed to perform a particular
arithmetic task or set of tasks.
• Skilled engineers had to manipulate parts of the computer’s
hardware directly.
– Some computers required input via relay switches
• Engineer needed to position electrical relay switches manually.
– Others required programs to be hardwired.
• Hardwiring: Using solder to create circuit boards with
connections needed to perform a specific task.
 The Programming
Language Continuum
 In the beginning… To use a computer, you needed to know how to program
it.
 Today… People no longer need to know how to program in order to use the
computer.
 To see how this was accomplished, lets investigate how programming
languages evolved.
• First Generation - Machine Language (code)
• Second Generation - Assembly Language
• Third Generation - People-Oriented Programming Languages
• Fourth Generation - Non-Procedural Languages
• Fifth Generation - Natural Languages
 The Programming
Language Continuum
 First Generation - Machine Language (code)
• Machine language programs were made up of instructions written in
binary code.
– This is the “native” language of the computer.
– Each instruction had two parts: Operation code, Operand
• Operation code (Opcode): The command part of a computer
instruction.
• Operand: The address of a specific location in the computer’s
memory.
– Hardware dependent: Could be performed by only one type of
computer with a particular CPU.
 The Programming
Language Continuum
 Second Generation - Assembly Language
• Assembly language programs are made up of instructions written in
mnemonics.
READ num1 » Mnemonics: Uses convenient alphabetic abbreviations to
READ num2 represent operation codes, and abstract symbols to represent
LOAD num1 operands.
ADD num2 » Each instruction had two parts: Operation code, Operand
STORE sum » Hardware dependent.
PRINT sum
» Because programs are not written in 1s and 0s, the computer
STOP
must first translate the program before it can be executed.
 The Programming
Language Continuum
 Third Generation - People-Oriented Programs
• Instructions in these languages are called statements.
– High-level languages: Use statements that resemble English phrases
combined with mathematical terms needed to express the problem or
task being programmed.
– Transportable: NOT-Hardware dependent.
– Because programs are not written in 1s and 0s, the computer must
first translate the program before it can be executed.

• Examples: COBOL, FORTRAN, Basic (old version not new), Pascal, C

 The Programming
Language Continuum
 Pascal Example: Read in two numbers, add them, and print them
out.

Program sum2(input,output);
var
num1,num2,sum : integer;

begin
read(num1,num2);
sum:=num1+num2;
writeln(sum)
end.
 The Programming
Language Continuum
 Fourth Generation - Non-Procedural Languages
• Programming-like systems aimed at simplifying the programmers task of
imparting instructions to a computer.
• Many are associated with specific application packages.
– Query Languages:
– Report Writers:
– Application Generators:

– For example, the Microsoft Office suite supports macros and ways to
generate reports
 The Programming
Language Continuum
 Fourth Generation - Non-Procedural Languages (cont.)
• Object-Oriented Languages: A language that expresses a computer
problem as a series of objects a system contains, the behaviors of those
objects, and how the objects interact with each other.
– Object: Any entity contained within a system.
• Examples:
» A window on your screen.
» A list of names you wish to organize.
» An entity that is made up of individual parts.
– Some popular examples: C++, Java, Smalltalk, Eiffel.
 The Programming
Language Continuum
 Fifth Generation - Natural Languages
• Natural-Language: Languages that use ordinary conversation in one’s
own language.
– Research and experimentation toward this goal is being done.
• Intelligent compilers are now being developed to translate
natural language (spoken) programs into structured machine-
coded instructions that can be executed by computers.
• Effortless, error-free natural language programs are still some
distance into the future.
 Assembled, Compiled, or
Interpreted Languages
 All programs must be translated before their instructions can be
executed.
 Computer languages can be grouped according to which
translation process is used to convert the instructions into binary
code:
• Assemblers
• Interpreters
• Compilers
 Assembled, Compiled, or
Interpreted Languages
 Assembled languages:
• Assembler: a program used to translate Assembly language programs.
• Produces one line of binary code per original program statement.
– The entire program is assembled before the program is sent to the
computer for execution.
– Similar to the machine code exercise we did in class
• Example of 6502 assembly language and machine code:
– JSR SWAP 20 1C 1F
– LDA X2 A5 04
– LDY =$80 A0 80
– STY X2 49 80
 Assembled, Compiled, or
Interpreted Languages
 Interpreted Languages:
• Interpreter: A program used to translate high-level programs.
• Translates one line of the program into binary code at a time:
• An instruction is fetched from the original source code.
• The Interpreter checks the single instruction for errors. (If an
error is found, translation and execution ceases. Otherwise…)
• The instruction is translated into binary code.
• The binary coded instruction is executed.
• The fetch and execute process repeats for the entire program.

• Examples: Lisp, Prolog, Java, JavaScript (used on Web Pages)

 Interpreted Programs

Machine Language
Source Code Statement
X=3 Interpreter 11011101
X=X+1
…

Next statement Execute

 Assembled, Compiled, or
Interpreted Languages
 Compiled languages:
• Compiler: a program used to translate high-level programs.
• Translates the entire program into binary code before anything is sent to
the CPU for execution.
– The translation process for a compiled program:
• First, the Compiler checks the entire program for syntax errors in the
original source code.
• Next, it translates all of the instructions into binary code.
» Two versions of the same program exist: the original source code
version, and the binary code version (object code).
• Last, the CPU attempts execution only after the programmer requests that
the program be executed.
• Examples: C, C++, C#, Java, Pascal, Visual Basic
 Assembly/Compiling Process
Human Brain English Algorithm

High Level Language Low Level Language -

(C++, C, Pascal…) Assembly

compiler assembler

Machine Code

If there are multiple source files that make up a final program,

these source programs must then be linked to produce a final
executable.
 Compilers
 Compilers on different machines generally produce different machine code,
targeted for that specific system.
• Mac and PC machine code different, can’t execute programs compiled for the
other

PC Compiler PC Machine Code

C++ source
Mac Compiler Mac Machine Code

• Note that under this model, compilation and execution are two different
processes. During compilation, the compiler program runs and translates source
code into machine code and finally into an executable program. The compiler
then exits. During execution, the compiled program is loaded from disk into
primary memory and then executed.
 Interpreted vs. Compiled
 What happens if you modify the source on a compiled programming
language (without recompiling) vs. an interpreted programming language and
execute it?
 Compiled
• Runs faster
• Typically has more capabilities
– Optimize
– More instructions available
• Best choice for complex, large programs that need to be fast
 Interpreted
• Slower, often easier to develop
• Allows runtime flexibility (e.g. self-modifying programs, memory management)
• Some are designed for the web
 Java?
 The astute members of the audience might have noticed that
Java was listed under both Interpreted and Compiled!
 A Java compiler translates source code into machine
independent “byte code” that can be executed by the java
“virtual machine”.
• Java Virtual machine doesn’t actually exist – it is simply a specification
of how a machine would operate if it did exist in terms of what machine
code it understands.
• Interpreters must then be written on the different architectures that can
understand the virtual machine and convert it to the native machine code
Java
compiler Mac Interpreter
Public class Foo {
01010001
if (e.target=xyz) then PC Interpreter
01010010
this.hide();
} PalmPilot Interpreter
 Java Benefits
 The great benefit of Java is that if someone (e.g. Sun) can write
interpreters of java byte code for different platforms, then code
can be compiled once and then run on any other type of
machine.
• No more hassles of developing different code for different platforms
 Sound too good to be true?
• Unfortunately there is still a bit of variability among Java interpreters, so
some programs will operate differently on different platforms.
• The goal is to have a single uniform byte code that can run on any
arbitrary type of machine architecture
• Java programs, due to the interpreted nature, are also much slower than
native programs (e.g., those written in C++)
 Building a Program
 Whatever type of problem needs to be solved, a careful thought out plan of
attack, called an algorithm, is needed before a computer solution can be
determined.

1) Developing the algorithm.

2) Writing the program.
3) Documenting the program.
4) Testing and debugging the program.

The danger is to jump straight to writing the code without thinking

about how to solve the problem first!
 Building a Program
 1) Developing the algorithm.
• Algorithm: A detailed description of the exact methods used for solving
a particular problem.
• To develop the algorithm, the programmer needs to ask:
– What data has to be fed into the computer?
– What information do I want to get out of the computer?
– Logic: Planning the processing of the program. It contains the
instructions that cause the input data to be turned into the desired
output data.
 Building a Program
 A step-by-step program plan is created during the planning
stage.
 The three major notations for planning detailed algorithms:
• Flowchart: Series of visual symbols representing the logical flow of a
program.
• Nassi-Schneidermann charts: Uses specific shapes and symbols to
represent different types of program statements.
• Pseudocode: A verbal shorthand method that closely resembles a
programming language, but does not have to follow a rigid syntax
structure.
 Building a Program
Flow chart: Nassi-Schneidermann chart:
Start Y If money > $10.00 N
Go out Go home
Count Money Repeat until
money < $10.00
Stop
Do you Yes
have more than Go out Pseudocode:
$10.00? 1. If money < $10.00 then go home
No Else Go out
2. Count money
Go home
3. Go to number 1
End
 Example Impact of Algorithms
 Searching a sorted list of names for some target name
• E.g. looking up a phone number for someone
 First algorithm: linear search
• Compare first name in the list
• If it matches, return match, otherwise continue with the next name in the list
• This works fine, but is inefficient for very large lists
 Second algorithm : binary search
• Start in the middle of the list
• If target name = name in the middle, return match
• If target name < name in the middle, repeat process on first half of the list
• If target name > name in the middle, repeat process on second half of the list

• Eliminates half of the list each time, much faster than linear search for long lists (lg N vs. N
for a list with N names)
 Algorithm can have a huge impact on efficiency and ease of implementation for the
solution!
 Building a Program
 2) Writing the Program
• If analysis and planning have been thoroughly done, translating the plan
into a programming language should be a quick and easy task.

 3) Documenting the Program

• During both the algorithm development and program writing stages,
explanations called documentation are added to the code.
– Helps users as well as programmers understand the exact processes to
be performed.
 Building a Program
 4) Testing and Debugging the Program.
• The program must be free of syntax errors.
• The program must be free of logic errors.
• The program must be reliable. (produces correct results)
• The program must be robust. (able to detect execution errors)

• Alpha testing: Testing within the company.

• Beta testing: Testing under a wider set of conditions using
“sophisticated” users from outside the company.
 Software Development:
A Broader View
Measures of effort spent on real-life programs:
Comparing programs by size:

Type of program Number of Lines

The compiler for a language with a
limited instruction set. Tens of thousands of lines
A full-featured word processor. Hundreds of thousands of lines
A microcomputer operating system. Approximately 2,000,000 lines
A military weapon management program.
(controlling missiles, for example) Several million lines
 Software Development:
A Broader View
• Measures of effort spent on real-life programs: Comparing
programs by time:
• Commercial software is seldom written by individuals.
– Person-months - equivalent to one person working forty hours a
week for four weeks.
– Person-years - equivalent to one person working for twelve months.

– Team of 5 working 40 hours for 8 weeks = ten person-months.

 Much more on these issues in the software engineering course

 Short History of PL’s
 1958: Algol defined, the first high-level structured language with a systematic syntax.
Lacked data types. FORTRAN was one of the reasons Algol was invented, as IBM
owned FORTRAN and the international committee wanted a new universal language.
  1965: Multics – Multiplexed Information and Computing Service. Honeywell
mainframe timesharing OS. Precursor to Unix.
 1969: Unix – OS for DEC PDP-7, Written in BCPL (Basic Combined Programming
Language) and B by Ken Thompson at Bell Labs, with lots of assembly language.
You can think of B as being similar to C, but without types (which we will discuss
later).
 1970: Pascal designated as a successor to Algol, defined by Niklaus Wirth at ETH in
Zurich. Very formal, structured, well-defined language.
 1970’s: Ada programming language developed by Dept. of Defense. Based initially
on Pascal. Powerful, but complicated programming language.
 1972: Dennis Ritchie at Bell Labs creates C, successor to B, Unix ported to C.
“Modern C” was complete by 1973.
 Short History of PL’s
 1978: Kernighan & Ritchie publish “Programming in C”, growth and popularity
mirror the growth of Unix systems.
  1979: Bjarne Stroustrup at Bell Labs begins work on C++. Note that the name
“D” was avoided! C++ was selected as somewhat of a humorous name, since “+
+” is an operator in the C programming language to increment a value by one.
Therefore this name suggests an enhanced or incremented version of C. C++
contains added features for object-oriented programming and data abstraction.
  1983: Various versions of C emerge, and ANSI C work begins.
  1989: ANSI and Standard C library. Use of Pascal declining.
  1998: ANSI and Standard C++ adopted.
  1995: Java goes public, which some people regard as the successor to C++.
Began as “Oak” within Sun.
 2001: Under development: C# (C-Sharp), language promoted by Microsoft with
similarities between C, C++, Java, and Visual Basic