I LL I NOIS I N S T I T U T E
OF T ECH NOLOGY
Transforming Lives. Inventing the Future.
[Link]
SOFTWARE ENGINEERING
CS 487
Prof. Dennis Hood
Computer Science
Lecture 2
Software
Processes
Lesson Overview
• Software Processes and Agile Methods
• Reading
– Ch. 2-3
• Objectives
– Explore current software engineering approaches
– Discuss the step-by-step approach to engineering
software systems
– Examine the roles of people and technology in the
development and use of software systems
– Analyze Agile methods and related approaches
Participation – P1
• Identify a real-world “gate keeping” scenario
– Explain how the gate impedes flow and why
– Describe design elements which maximize flow
– Describe the “authority” which enforces
compliance
• Describe one benefit and one drawback of
allowing agility in the software development
process
• Explain the value of “surfacing” in an iterative
development process
Software
Processes
Software Process Models
• Software process
– A set of activities that leads to the production of a
software product
– e.g., analysis, design, build, verification
• Generic models
– Process frameworks meant to be tailored
• Waterfall
• Incremental Development
• Reuse-Oriented SW Engineering
Waterfall
• Cascading from one phase to the next
• Each phase ends with a “gate”
– Signoff is required before a phase can be
considered complete
– Increases the likelihood of true completion
– Makes it difficult to estimate completion time
• In practice, any engineering effort follows this
general approach, but also involves iteration
– Iterations facilitate learning
– It is difficult to estimate the required number of
iterations
Iterative Development
• Achieve a defined “sub-objective” with each
iteration
• The product is evolved methodically
• The focus is on answering questions and
resolving challenges
• Often used to bridge the project team-
customer gap
– Produce a prototype and solicit feedback
– Quickly and cheaply achieve critical understanding
Reuse-Oriented SW Engineering
• Focus is on reusing previously created system
elements
– Clearly faster
– Defects are limited to points of integration and
“inappropriately” applied components
• Tradeoffs may be necessary
– Existing components may not exactly match
requirements
• Requires disciplined management of reusable
components
– Documentation
– Testing
– Version control, etc.
Development Process Objectives
• Each of these models addresses the following
needs:
– Specification of requirements
• Identifying services to be provided and environmental
constraints
– Design and implementation
• Transforming specification into working system
– Verification and validation
• Showing that the system satisfies requirements and
meets customer expectations
– Evolution
• Growing the system over time to accommodate
environmental changes and new requirements
The Spiral Model
• Iterative, risk-driven approach developed by
Barry Boehm
• Four phases executed repeatedly
– Objective setting
– Risk assessment and reduction
– Development and validation
– Planning for the next iteration
The Rational Unified Process
• Hybrid model incorporating prototyping and
incremental delivery
– Based on UML and the Unified Software
Development Process
• Described from multiple perspectives
– Dynamic: phases of the model over time
– Static: process activities that are enacted
– Practice: suggests good practices to be used during
the process
• RUP phases:
– Inception, elaboration, construction, and transition
RUP Best Practices
• Develop software iteratively
• Manage requirements
• Use component-based architectures
• Visually model software
• Verify software quality
• Control changes to software
Change Happens
• An iterative approach is beneficial because
change is likely
– Technology evolves rapidly
– Competition drives customers to change
requirements
– Changes to economic conditions can affect
strategic approaches
• Minimizing time and scope also minimizes
opportunity for change
– Plan for relatively short, well-defined iterations
– Customers can utilize incrementally delivered
systems
Agile
Methods
Agile Methods
• Software development evolving to keep up
with the dynamic world it serves
– Speed up delivery to hit the moving target
– Involve users significantly and early
– Extreme flexibility
• Agile characteristics
– Interleaving of phases
– Documentation effort is minimized (focus on
essentials)
– Frequent delivery of versions
– Visual development tools to facilitate rapid,
interactive UI development
The Agile Approach
• Flow
– Get a basic idea, then jump into design and
implementation cycles
• Requirements gathering and V&V are pulled into design
and implementation
– Frequently “surface” the evolving product
– Growth is more organic than planned
• Pros and Cons
– The benefit is that progress drives discovery which
drives more progress
– The risk is the discovery is influenced by the
working prototype and presence of the customer
Extreme Programming
• Best practices pushed to the extreme
– Extremely rapid development and deployment
– Extremely small teams
– Extremely close user involvement
– Extremely tight iterations
• Requirements come as simple customer stories
(scenarios)
• Change is handled by frequently coding, testing
and releasing
• Simplicity is a goal to help manage the rapid
pace
XP Testing
• The lack of a system specification makes it
difficult to “hand off” testing to an
independent verification team
• Focus is on only bringing “clean” code into the
current version
– Develop the tests first, then the code
– In some ways developing the tests is very similar to
specifying functionality, designing the interface,
etc.
– Especially if the user writes the tests
– Rely on automation
Agile Project Management
• Project management has a generally
bureaucratic feel to it
• Agile methods require more flexibility and
speed than what is normally associated with
project management
• The Scrum approach is an example of PM
principles applied to short, tight iterations
– Short, fixed timeframes (2-4 weeks)
– Focus on prioritized, outstanding work
– Frequent communication and assessment
– Commitment to iterations
Scaling Agile Methods
• Agile methods seem naturally suited for small
development efforts
• That would be disappointingly limiting
• Scaling to larger development is possible with:
– Focus on up-front analysis and design
– Cross-team communication mechanisms
– Continuous integration including whole-system
builds every time anything is changed/added
– PMs with experience with Agile methods
– Organizations with a certain degree of flexibility
