MurruDeiasMugheddu03

Orlando Murru & Roberto Deias & Giampiero Mugheddu
Assessing XP at a European Internet Company
IEEE Software Magazine V20n3(Apr/May 2003)pp37-43
=EXPERIENCES Fst EURO PARTNER SECURE PORTALS RFCs JAVA RUP XP
Table (p38) shows 2 projects. 6/12 adoption, 2 9/12 adoption.
No planning means no control and coordination.
Planning: involve all, separate estimates from rewards and designs.
Metaphor? Felt need for architectural planning. (no stand up meetings?). Used posters and sketches instead.
refactoring and testing needs experience (didn't rotate?)
Sales and management had problems.
XP can imply ISO9001!

WoodKleb03

William A Wood & William L Kleb
Exploring XP for Scientific Research
IEEE Software Magazine V20n3(Apr/May 2003)pp30-36
=EXPERIENCE XP SCIENTIFIC NUMERICAL AEROSPACE NASA RUBY CULTURE
table (p33) shows 10/12 practices full adopted.
Science culture clashes with XP: TSP/PSP vs iteration, ISO9001 vs no specs, individual rewards vs team ownership, delay before acceptance, room layout.
2 people. One member of team had to play the customer.
Numerical methods thought to need early optimization not simplest design. Wrote stories to separately find and fix hot spots. So, fixed the right problems. Rediscovered Jackson's Optimization: Don't & Don't Yet Law.
Pair programming worked.
Will Collective ownership hinder individual promotion?
xUnit testing worked. But added timing and numerical stress (accuracy) tests.
Planning game helped develop increments. Weak plan - slower velocity.
2kloc at 27line/hour, 912 lines of production + 1k test code + 0.5K for testing scripts ad utilities.
Better quality code: more readable.

Jonathon Rasmusson
Introducing XP into Greenfield Projects: Lessons Learned
IEEE Software Magazine V20n3(Apr/May 2003)pp21-28
=EXPERIENCE ThoughtWorks XP Java WEB SERVICE ENERGY
21KLoC + 16KLoTESTING
Describes 11 practices adopted. 2 week cycles.
The little things (snacks) count. Separate offices slowed things down.
Patterns vs XP? Make it work first, refactoring will uncover the right patterns.
Testing became 24mins and too long.... needed MockObject perhaps?
daily meetings: set high expectations at the start!
test-infected.
TDD:="Test Driven Development" -- is not easy at first, easy to slide back to code first, test after.
Problem: working with other traditional technical groups: need extra communication.
Strive for a single business customer.
Integrate testers (QA) into the team.
Table (p27): partial implementation of metaphor, TDD, and on-site customer.

Laurie Williams
The XP Programmer: The Few-Minutes Programmer
IEEE Software Magazine V20n3(Apr/May 2003)pp16-20
=EXPERIENCES XP
Describes XP practices.
Compares XP to "The One-Minute Manager": rapid consistent simple feedback -- from requirement to test to code to client very rapidly.
Save time on big early phases by keeping them small and use the saved time to incrementally grow to what the user wants, even as the user learns what they need and moves the target...
anecdotes claim that growing a big working system one story at a time costs no more than trying to get the whole thing right first. Research in progress.

Donald J Reifer
XP and the CMM
IEEE Software Magazine V20n3(Apr/May 2003)pp14-15
=EXPERIENCE XP SW-CMM PROCESS SQA TESTING SCHEDULES
In theory [Paulk01] XP and CMM level 2&3 fit. In two real projects problems developed. Both projects have to handle bad schedules and novelty.
(dick) |- Neither implemented XP properly.
In one case the CMM group argues that their process is already the best practice and the XP group argues they don't have time for CMM practices.
In the other case the XP people sacrificed test-first programming and are being criticized by the QA team. QA says the quality is down, the XP manager says QA is adding no value to the software.
Need SEI to develop and advertise a mapping of CMM onto XP as a level 5 change management example perhaps.