XP is the most important movement in our field today.
-- Tom DeMarco
Extreme Programming (XP) is an agile software-development methodology.
XP helps you remain light on your feet by avoiding unnecessary
baggage and by incorporating feedback continuously. Changing
requirements are an expected and acceptable risk, because the customer
sees the system being developed in real-time. Mistakes are
immediately visible and are corrected while the feature's
implementation is fresh and pliable, much like a potter reworks clay.
Programmers work and rework the code in XP projects. The customer
sees a system grow from layer upon layer of detail. The software is
only as effective as the details it embodies. A tax accounting system
must round computations correctly, or it can't be used; it's
insufficient to get the formula right without considering that taxes
are collected in whole currency units. Details matter, and XP
programmers reflect back to the customer in the only way that
matters: working code.
All this working and reworking requires a stable base and good tools.
To throw pots effectively, you need to be seated comfortably at your
potter's wheel, and your tools need to be within easy reach. In the
world of idea creation, people need comfort, too. They need to know
what's expected of them and why. An XP team is expected to follow 12
simple practices. You aren't supposed to execute the practices
blindly, however. XP gives us a framework of four core values that
lets us adjust the practices to suit our particular project. The four
core values are like a comfortable mental chair; we work our code
using the practices with the core values supporting our every
movement.
This chapter explains XP's core values: communication,
simplicity, feedback, and courage. We then enumerate XP's 12
practices, and discuss how to adopt them.
Core Values
XP is built on four core values: communication, simplicity, feedback,
and courage. The values reinforce each other to form a stable
structure as shown in the figure:
Core Values
The four values give the people in XP projects a firm foundation to
stand on, the why for the how. Unlike plan-driven software
methodologies mentioned in The Problem, XP is
people-driven. We value people over process.
The idea of giving people reasons (core values) for what they do
(practices) is not new. For example, before XP was a twinkle in Kent
Beck's eye, John Young, then CEO of Hewlett-Packard, stated,
"We distinguish between core values and practices; the core
values don't change, but the practices might."[1]
It's important to trust people to judge the validity of a practice for
their particular job. They need a value system to frame their
judgments so that a person can change a practice without undermining
the goals of an organization.
The values relate to each other to form a framework. Without these
relationships, the values would not hold together, and people would be
less likely to accept and to work with them. The tetrahedron
symbolizes the importance of the bonds between the values. As you
read through the descriptions in the following sections, you'll see
how the values support each other and the practices.
Communication
A little Consideration, a little Thought for Others, makes all the difference.
-- Eeyore (A. A. Milne)
Software is developed as quickly as the communication links in the
project allow.
The customer communicates her requirements to programmers. The
programmers communicate their interpretation of the requirements to
the computer. The computer communicates with its users. The users
communicate their satisfaction with the software to the customer.
Communication in XP is bidirectional and is based on a system of small feedback
loops. The customer asks the users what they want. The programmers
explain technical difficulties and ask questions about the
requirements. The computer notifies the programmers of program errors
and test results.
In an XP project, the communication rules are simple: all channels
are open at all times. The customer is free to talk to the
programmers. Programmers talk to the customer and users. Unfettered
communication mitigates project risk by reducing
false expectations.
All stakeholders know what they can expect from the rest of the team.
Simplicity
Pooh hasn't much Brain, but he never comes to any harm. He does silly
things and they turn out right.
-- Piglet (A. A. Milne)
We all want simple designs and simple implementations, but simple is
an abstract concept, difficult to attain in the face of complexities.
XP takes simplicity to the extreme with practical
guidelines:
Do the simplest thing that could possibly work (DTSTTCPW),
Represent concepts once and only once (OAOO),
You aren't going to need it (YAGNI), and
Remove unused function.
Do the simplest thing that could possibly work (DTSTTCPW) means you
implement the first idea that comes to mind. This can be scary. Rely
on your courage to try out the idea. Remember that failure is
an important part of creation. It is unlikely the simplest idea is what you
will end up with. However, it's also unlikely you can anticipate
what's wrong with your simple solution until you try it out. Let the feedback system guide
your implementation.
DTSTTCPW is simplicity as in fast and easy.
Once and only once (OAOO) helps you maintain your agility by reducing
the size of your code base. If you let
conceptual redundancy permeate your system, you have to spend more and
more time rooting out faults. Every time you copy-and-paste, you
take one more step closer to bloatware. Each copy creates an
implicit coupling, which must be communicated to the rest of the team.
Be courageous, just say no to your mouse. Say yes to refactoring the
code for re-use. OAOO is simplicity as in few interchangeable parts.
You aren't going to need it (YAGNI) is a popular and fun expletive.
If you can solve the
immediate problem without introducing some feature, that's YAGNI! And,
you simplified your problem by omission. YAGNI is a corollary of
OAOO. If you don't have to implement the feature in the first place,
your system just took a step away from bloatware.
YAGNI is simplicity as in basic.
Sometimes you add a function for good reason but later find out the
reason is no longer valid. At this point you should delete the
function. It is unnecessary complexity. It shouldn't require much
courage, because the code is still there in your source repository.
[COMMENT: link to logistics.]
You can always pull it out if you need it again.
Removing dead code is simplicity as in pure and uncluttered.
Feedback
Well, either a tail is or isn't there. You can't make a
mistake about it. And yours isn't there!
-- Pooh (A. A. Milne)
The more immediate feedback, the more efficiently a system functions.
A simple example can be found in the shower. Some showers respond
instantly to changes in the faucet handle. Other showers don't. I'm
sure you've experienced showers installed by Central Services engineers
from the movie Brazil.[2] You turn on the shower, adjust the temperature,
and hop into a hailstorm or The Towering
Inferno.[3]
After you peel yourself off the shower wall, you adjust the temperature,
and wait a bit longer before timidly stepping in again.
The long delay in the system makes showering unpleasant and
inefficient.
For many customers, this is what software development is like.
You request a change, and it is delivered many months later in some big
release. Often the change fails to meet your expectations, which
means another change request with yet another long delay.
XP is like a well-designed shower. You request a change and out comes
software. Adjustments are visible immediately. The customer sees her
requirements or corrections implemented within weeks. Programmers
integrate their changes every few hours, and receive code reviews and
test results every few minutes. Users see new versions every
month or two.[4]
The value of immediate, real world feedback should not be
underestimated. One of the reasons for the success of the Web is the
abundance of structured and immediate feedback from users. Developers
see errors in real time, and contract all input and output that causes
Web application failures. Users benefit from running the latest
version of the software, and seemingly on demand fault corrections.
When people talk about the enduring value of the Web in the distant
future, I predict they will value the extreme acceleration of user
feedback and software releases. The impact of this feedback on
quality and development efficiency is what differentiates Web
applications.
XP reduces project risk by taking iterative development to the
extreme. The customer's involvement does not end at the planning
phase, so requirements errors are reconciled almost immediately. The
system's internal quality is maintained by programmers working in
pairs who are striving for simplicity. Automated testing gives
everybody feedback on how well the system is meeting expectations.
XP uses feedback to integrate
towards a solution, rather than trying to get it through
a discontinuity.[5]
Courage
It is hard to be brave, when you're only a Very Small Animal.
-- Piglet (A. A. Milne)
Fear is a prime motivator, or as Napoleon Bonaparte put it, "There
are two levers for moving men: interest and fear."
With courage, our relationships take on a new quality: trust. XP helps
build the bonds of trust by repeatedly exposing people to small successes.
Courage is required at all levels. Is this solution too simple? Is it
too complex?
Does this test cover all the cases which could possibly break?
Will the programmers understand what I mean by the story? Will we
make
it to
Comdex without a detailed schedule?
We overcome fear, uncertainty, and doubt in XP with courage backed
by the other three values. A simple system is harder to break than a
complex one. Multilevel, rapid feedback lets us know quickly when our
courageous changes fail.
Open communication means we don't have to face our fears alone. Our
team members will support us. All we have to do is speak of our fears
as openly as Piglet did in the epigraph to this section.
And, Rabbit finds the right words to support him[6]:
"It is because you are a very small animal that you will be Useful in
the adventure before us."
Piglet was so excited at the idea of being Useful that he forgot to be
frightened any more [...] he could hardly sit still, he was so eager
to begin being useful at once.
Sometimes we feel as small and ineffectual as Piglet. During these
downtimes, it's likely one or more of our team members feel as
courageous as Rabbit or Pooh.
XP accepts that people's emotions vary, so XP uses team interactions
to keep the project stable and to provide emotional support in those
inevitable, difficult times.
Courage is a double-edged sword. You needed to overcome your fears,
but too much courage can be dangerous. XP uses small steps to promote
courage and keep it in check. Team members see a continuous flow of
failures and successes. XP uses small, regulated doses to discourage
excess and encourage success.
The Practices
XP's practices embody the values described in the previous sections.
In his book Extreme
Programming Explained Kent Beck defines the 12 practices
as follows (quoted verbatim):
- The Planning Game
Quickly determine the scope of the next release by combining business
priorities and technical estimates. As reality overtakes the plan,
update the plan.
- Small releases
Put a simple system into production quickly, then release new
versions on a very short cycle.
- Metaphor
Guide all development with a simple shared story of how the whole
system works.
- Simple design
The system should be designed as simply as possible at any given
moment. Extra complexity is removed as soon as it is discovered.
- Testing
Programmers continually write unit tests, which must run flawlessly
for development to continue. Customers write tests demonstrating the
features are finished.
- Refactoring
Programmers restructure the system without changing its behavior to
remove duplication, improve communication, simplify, or add flexibility.
- Pair programming
All production code is written with two programmers at one machine.
- Collective ownership
Anyone can change any code anywhere in the system at any time.
- Continuous integration
Integrate and build the system many times a day, every time a task is
completed.
- 40-hour week
Work no more than 40 hours a week as a rule.
Never work overtime a second week in a row.
- On-site customer
Include a real, live user on the team, available full-time to answer
questions.
- Coding standards
Programmers write all code in accordance with rules emphasizing
communication through the code.
These 12 simple practices realize the four core values.
The remainder of this book explains how to implement the practices in
detail. Before we get into implementation, let's briefly discusses
how you might adopt XP in your organization.
Adopting XP
Organizations have their own way of doing things. There are
practices, processes and probably even some core values. In order to
adopt XP, you'll need to work within this framework, and accept the
status quo. One way to start is to use XP in a project composed of
volunteers. The project may even be important, which is good, because
your success with XP should be visible. Once that project succeeds, pick
another and let another team coalesce around it, possibly including a
few members, but not all, from the original XP team.
You may not be in a position to pick and choose projects, but you
probably have some latitude in how you work on a daily basis.
If this is the case, try selecting a few practices that align with
your existing methodology. For example, if your organization values
testing, try test first programming. Or, organize your day around
stories. I find this technique to be helpful for non-software
problems, too, as you'll see in Release Planning.
Keep the stories on index cards, and work through them serially. Check off
each card as you complete it.
Once you see your productivity go up, discuss the practices you found
successful with a co-worker or your manager. Use the core values as
your guide. You'll need courage to start the communication. Keep
your explanations simple, focusing on the practice, not the whole of
XP. Be open to feedback, and incorporate it immediately. And, as
always in XP, take small steps and iterate.
As you read through this book, look at the practices from your
organization's perspective. You'll see plenty of ways to integrate
them. XP is an evolutionary methodology that can be adopted
incrementally and organically. Even if you are the head honcho,
don't install XP with a Big Bang. Organizations and people have their own
natural pace of change. XP cannot accelerate this rate of change
overnight.
Change cannot be mandated. Rather XP values feedback and
communication to allow you to measure your progress and to integrate
change continuously.
Footnotes
As cited in
Built to Last, Jim Collins and Jerry Porras,
HarperBusiness. 1997, p. 46.
http://www.filmsite.org/braz.html
http://www.norcalmovies.com/TheToweringInferno
-
XP's founders recommend multi-week iterations and releases to the
customer every three or four iterations. My experience with Extreme
Perl is that an iteration per week and one iteration per release works
well, too. See Logistics for how to make this happen.
Thanks to Johannes Rukkers for this excellent observation.
All A. A. Milne quotes in this chapter are from
Winnie-the-Pooh,
A. A. Milne,
Dutton's Childrens Books,
1994.
|