Lecture 3: Software Engineering.
Overview
q Analytical and Algorithmic methods address problems
that are specific, concise and not too complicated
q Typical real-life problems are large-scale in nature
and require a team effort to design and implement
q SE method addresses these more complicated real-life
problems
Objectives of this lecture
q Introduce the field of Software Engineering
q Discuss the basic software lifecycle and what is
required at each stage of the cycle
--the waterfall model
What is & Why Software Engineering?
q A computer's behaviour is completely dictated by
software.
q In order to make efficient and effective use of
computers, we must design software systems that are reliable, flexible,
expandable, efficient, and verifiable
q In the late 1950s and early 1960s software systems
were generally built without following any established, systematic methodology
q This lead to a serious crisis as every serious
modification in needs, involved re-writing the whole software.
q Experiences in the analysis of previous software
development failures forced software designers to adopt more scientific
approach to software development
q Software Engineering (SE), therefore, is an attempt to
apply the structured methods of engineering to software development
q It involves the analysis and application of strategies
to design, construct, and maintain large-scale software systems using certain
common methods.
q The goal of SE is to ensure the development of
software that has the characteristics listed in the second point above
q SE emphasizes on
Ø
Conceptual
understanding: requires the detailed understanding of the requirements of the
system to be implemented in software
Ø
Design:
successive refinement of abstract models toward concrete implementation
Ø
Prototyping:
construction trial model or preliminary models to be carefully examined before
building the final system
Algorithm Design Tools
q Hierarchy chart
Ø
Used to
show, pictorially, the phases of software design and execution in terms of
levels of increasing details
q Program documentation
Ø
Gives a
clear statement of the purpose and design of a piece of software
Ø
Should include
at least, the title, description, formal statement, author & date
Ø
Helps
programmers involved in a large SE project to understand each other's programs
Ø
Although
useful, documentation should be used carefully so that the source program is
not buried inside the documentation
Algorithm Design Strategies
There are two main algorithm
design approaches namely:
q Bottom-up
Ø
This
approach identifies and constructs the basic building blocks of a large system,
then incrementally builds larger subsystems using smaller ones, until the whole
system desired is constructed
Ø
This
approach may be useful in projects such as creating software library or
scientific research.
Ø
The method
is generally bad when constructing software whole requirements can be
pre-determined.
q Top-down
Ø
Requires a
more detailed understanding of the problem at hand
Ø
Overall
problem is then broken down into smaller logic subunits
Ø
Specifications,
hierarchy charts, and algorithms are then developed
Ø
Must
understand the design of all phases of the project before the first line code
is written.
Ø
This method
requires more preparatory work and thus too formal to be attractive to
beginners.
The Waterfall Model
q This model describes six phases found in the life
cycle of a professional software development process
q Needs Analysis:
Determines the customers’ needs
q Needs Specification:
Documents the needs in terms of input, process and output and verify
with the customer
q Algorithm Design:
Breaking the problem into modules using tools such as hierarchy charts
and develop algorithm for each module.
q Software implementation: Actual coding using a suitable language.
q Testing and Integration: Evaluation of the various
modules and integrating them to form the complete software.
q Delivery & Operation: Turning over the finish
product to the customer/installation
q Program Maintenance:
Could arise as a result of errors, changing needs or new
requirements. Could lead to starting
the project from any of the previous stages
--- life cycle
Criticisms of the Waterfall Approach:
q Real projects rarely follow the sequence – Iterations
always occur
q Difficult for the customer to specify all requirements
q The developer may be unsure of the efficiency of an
algorithm, etc.
q Working version is at the end – Customer must be
patient.
Other Paradigms
Prototyping:
q Enable the developer to create a model of the software
q It involves an Iteration of:
Ø
Requirement
gathering
Ø
Quick Design
Ø
Building
Prototype
Ø
Customer
evaluation
q It gives the customer confidence and allow the
developer to test his methods
q Problems:
Ø
Could also
lead to the customer losing confidence
Ø
The
developer often makes compromises to get the prototype working
The Spiral Method
q Combines the Classic life cycle and Prototyping
q It involves the sequence:
Ø
Planning
–determine objectives, alternatives, and constrains
Ø
Risk
analysis – analysis of alternatives and identification/ resolution of risks
Ø
Engineering
– development of the next level product
Ø
Customer
evaluation
q Currently the most realistic approach – although
relatively new
Fourth Generation
Technique (4GT)
q Involves software tools (4GL) that automatically
translates designs to products
q It involves the sequence:
Ø
Requirement
gathering
Ø
Design
Ø
Implementation
using 4GL
Ø
Testing
q Reduces time/cost
q Problems:
Ø
Current
tools are limited to business information systems