Case Studies Extreme Programming

1
Case Studies Extreme Programming

• Pradeep Krishnegowda

2
Agenda

• XP Refresher
• What is XP
• XP Practices
• Case Study eXpert System
• Introduction to the system
• Research Setting
• Technical Implementation Environment
• About the team
• Work environment
• Results
• Effort Distribution
• Discussion
• Summary

3
What is XP

• Extreme Programming is a discipline of software
  development based on values of simplicity,
  communication, feedback, and courage
• Follows 12 Core Practices

4
XP Practices
5
Case Study eXpert

• eXpert system for managing the research data
  obtained over years at a large Finnish Research
  Institute virtual file cabinet
• Development time and schedule were fixed
• Requirements not well known before the project
  was initiated

6
Research Setting

• Team of 4 developers/software engineers
• eXpert is a virtual file cabinet which holds a
  large number of organized rich links to physical
  or web based resources. The system is a web based
  client-server solution

7
Technical Implementation Environment
8
About the Team

• 5-6th year students with 1-4 years of industrial
  experience
• Well versed in java programming language,
  object-oriented analysis and design approaches
• Two day hands-on training on XP practices,
  development environment and software
  configuration management tools

9
Work Environment

• Worked in a co-located development environment
• Customer shared the same office space with
  development team
• Office space and workstations organized according
  to the suggestions made in the XP literature to
  support efficient team work

10
Results

• Precision time of 1 minute
• Development work size estimated by automated
  counters for Java and JSP
• Development time defects, post release defects
  and number of enhancement suggestions made by 17
  system testers
• Quality of data monitored by project manager,
  dedicated metrics and onsite customer

11
Results
12
Results Work Effort Hours
13
Results Team Productivity
14
Results Code Integration
15
Results User Stories
16
Results System Quality
17
Results System Quality
18
Results Customer Involvement
19
Results Rework Costs()
20
Effort Distribution Project Management
21
Effort Distribution Planning
22
Effort Distribution Coding
23
Effort Distribution Project Mettings
24

• User stories and tasks were documented and
  displayed on the walls of the development room
• System documentation for maintenance was produced
  during the last two releases when the system
  architecture and user functionality had
  stabilized enough and was not subject to constant
  changes.

25
Discussion

• Actual customer involvement in the project
• Customer was present close to 80 of the working
  time
• Only 21 of work effort used to assist
  development team
• presence of customer was a motivation for the
  development team

26
Post release defect density

• Was at an acceptable level, i.e. 1.43
  defects/KLoc
• Result can be seen positive from three
  perspectives
• First, an early insight was gained to the overall
  product quality
• Testing team forms a part of the user group who
  will make use of the system when it will be
  finally released
• Rapid feedback acquired by the development team

27
Pair Programming

• In the first release 81.7 of the programming
  effort was done in pairs. Reduced only slightly
  to an average 75.9 in the next releases
• Effort remained more than 70 shows that team
  felt comfortable with it
• Development time productivity achieved in the
  second release is close to the same as PSP
  research has consistently shown (W. Hayes and J.
  W. Over, "The Personal Software Process (PSP) An
  Empirical Study of the Impact of PSP on
  Individual Engineers," Software Engineering
  Institute, CMU/SEI-97-TR-001)

28

• A few user-stories had been postponed
• Not a surprise to the customer
• Finally, in the eXpert project 9.8 of the total
  development effort was rework costs

29
Summary

• The resulting product was tested by 17 testers
  who used a maximum of 45 minutes
• System release defect rate was 1.43 defects/Kloc
• Teams overall productivity was 16.90 Locs/hr
• Rework costs was a mere 9.8 of the total
  development effort
• Required onsite customer involvement was 21

30
QAs

• Thank you for listening
• Any questions?

31
References

• Proceedings of the 29th EUROMICRO Conference New
  Waves in System Architecture (EUROMICRO03)
• Abrahamsson, P., Koskela, J. (2004) Extreme
  programming Empirical results from a controlled
  case study,ACM-.IEEE International Symposium on
  Empirical Software Engineering (ISESE 2004),
  Redondo Beach CA, USA
• www.xprogramming.com
• http//groups.yahoo.com/group/extremeprogramming/