Review: Rapid Development

I just finished reading Steve McConnell's Rapid Development: Taming Wild Software Schedules. Well, many times it was not exactly "reading", more like "page-skipping". Obvious. Obvious. Obvious. But every time I thought I knew what was coming there was something unexpected hidden inside this big recapitulation of Fred Brook's findings. With big hard data pictures on the border. And then I'm horrified that I can relate to most of what is written, and that nothing's really new to me any more. Even if I didn't expect to read it from McConnell.

So the excitement of the days when I wrote stuff like Do You Understand XP? is gone. Rapid Development is a good book that reviews every practice it preaches with a critical eye, but I'm not overly excited. Yes, I learned, again, that many so-called Agile practices are known for a long time. But hey, I knew that after reading the Mythical Man Month. Excitement is hard to find in books these days. If I only could find one of those thought-provoking books, that rips your universe apart.

All Programmers Are Code Narcissists

01001101011110010010000001100011011011110110010001100101
00100000011100100111010101101100011001010111101000100001

I finally discovered the truth about why developers rather rewrite a 1MLOC project from scratch than trying to understand a fellow programmer's code:

We're all code narcissists!

And the reason for that can easily deducted in a tiny logical chain:
(The best code is easy to understand) ^ (I can understand my own code the easiest, duh!)
=> (My own code is the best code)

Unfortunately this is only true for me, or perfect clones of myself. Which rules out everybody else I work with. Which reminds me... What was the reason that programming is done in teams?

Test Everything That Could Possibly Break - A Guide To Better Testing

Joe: "Writing this test will make sure that we find bugs quicker. It will let us change the code without breaking anything and it will help us to write decoupled code."
Jim: "Maintaining this test will be a nightmare. It is tightly coupled to the class we're writing and we cannot change anything without changing the test. It will be a pain."
Joe: "How do you know?"
Jim: "Well, how do you know?"
Joe: "I have 20 years of experience not writing unit tests."

So what?
When I'm writing new code I am never sure whether I test enough, or if my tests are on the right level of abstraction. For complicated core functionality of a distributed system this is a no-brainer - I use TDD, which by the very definition gives me 100% code coverage, and add some nice integration and acceptance tests. But there are a myriad of cases where going forward in the baby-step TDD way seems a waste of time, and it would really help me to find some sensible rules to apply.

The simplest and best rule I have found so far is an idea from Extreme Programming:
Test Everything That Could Possibly Break

Unfortunately this rule is not as simple as it looks. My approach to that is the good old mantra of try-measure-adapt, where try means to just do whatever a randomly selected guy thinks is the new cool-aid, measure means to listen to that whimsical thoughts my brain produces while doing it and adapt means to look at the results and change my behavior.

So here's my guide to better testing, from the beginner to the professional level:

Beginner: Do some unimportant project by following the description of TDD step-by-step. Don't waste your employers money with that if you don't have any idea of how to do it - the first time you use it will be a disaster. Writing yet another Sudoku solver in your favorite programming language might be a cool idea.
The important part is that you don't have an idea of what "test everything that could possibly break" means, so your best bet is to assume everything might break. Even those getters and setters over there. Remember that you're not allowed to rant about made-up scenarios of why too many tests might be bad if you have never experienced what it means to maintain a program with too many tests. Do that first and come back later and read on.

Advanced: So you already have some experience doing TDD and know how it feels to write all those little unit tests. You got some feedback on when those tests caught a stupid bug that would have taken an hour to find if you hadn't written the test. You now know what the impact of unit tests on your ability to do refactorings is. Now go and break the rules by various degrees. Try to be less exhaustive with your tests and bundle your baby steps into bigger units of work. See how that affects your ability to find bugs. Test your assumptions and be aware of when they break. When you find a new bug that takes some hours to debug, think about what kind of test would have helped you find it quicker and write those tests from now on.

Expert: If I were an expert I could probably tell you more about what to do in that case. I still hope that repeating the advanced guidelines will finally make me as wise as Joel and Jeff in their discussions or Jay Fields when he writes about developer tests. Perhaps listening to those guys will enlighten you. You could even take a look at the very interesting discussion of the idea to test everything that could possibly break.

In a nutshell:

• Start by testing everything, even if it looks stupid (don't do it at work).


• Do slightly bigger steps and see what happens.


• Adapt whenever you experience a situation in which different behavior would have made more sense.

Leaving the Comfort Zone

A week ago I held my first talk since university. It was the first talk in my life I held in English, and I was scared like hell. It felt like living through those exam days back at school all over again - just that this time I thought I'm old enough to realize that it'll all be okay. I'm obviously not.

I acted just like I did back in school. Or worse. I waited for the last possible moment before I started preparing the talk. When I finally started with the slides, a little perfectionist devil Manuel sat down leisurely on my left shoulder, telling me that this crap is just not up to my own standards. I worked long hours the night before the big day, and woke up early just to be able to rehearse the whole play before entering the stage.

Of course everything worked out just fine. Well, besides me saying basic-a-lly all the time. So why was I so freaked out? Well, I was obviously leaving my comfort zone.

Now the interesting thing beneath all that personal drama I'm ranting about is that I suddenly realized for how long I did not really step out of my safe little comfort zone. Granted, applying at Google is not the most un-stressful experience I ever had. And obviously starting a new job with all those bright people around me did not exactly make me fell warm and cozy.

But at that very moment I stood there, my peers gazing absent-mindedly into their laptops, my hands slightly sweating, I suddenly realized that ever since I started working I was doing everything exactly in the way I was most comfortable with. And that was kind of a shock.

Of course after giving the talk I felt great. I had known that in advance, but I realized that without being nudged enough I'd probably have tried to wiggle out somehow.

Lessons learned:

1. It's easier to step out of the comfort zone when somebody kicks your ass.


2. It's a long way from leaving your comfort zone once to real change.


3. I must learn how to leave my comfort zone on my own.

Do you know tricks that make it easier to leave the comfort zone?

One Day In My Life As A Googler

Disclaimer: this is a personal entry. Which means that if you are here for technical revelations or to gather super secret information about Google, go away, or I'll bore you to death. Seriously. Don't tell me I didn't warn you.

6:00 am: The iPod station starts playing and lifts me gently from the land of dreams.
"... Don't run away the time is now the place is here ..."
Sasha's swinging voice finally makes me cautiously open my eyes: another sunny day in Mountain View. The weather gadget on my iPod informs me that the temperature is currently 11 degree Celsuis, probably rising to a comfortable 21 throughout the day. I realize that I still don't know how to convert Celsius to Farenheit without accessing the Internet.

6:40 am: Armed with a gbike and a helmet I finally leave the apartment, trying to get warmed up for my workout on the 3 mile ride to work. While I pass the sporadic early jogger I can see the first rays of sunlight painting the landscape in warm colors. Pushing the pedals gets my pulse up to 150. Sweet.

6:55 am: The gym is wonderfully empty at this time of the day. With just a handful of Googlers around me I enjoy a quite workout. Today I'm torturing my upper body.

7:45 am: I end my workout with an easy 50 minute run along Mountain View's beautiful shoreline. Squirrels pass my way while two chatting women overtake me. I realize that I've got a long way to go before I will become a decent runner. Well, at least I can program computers.

8:50 am: Breakfast time. I get myself some freshly made scrambled eggs with crispy bacon, two chocolate croissants, some healthy orange juice and a steaming cup of coffee. Throughout the next view minutes some of my colleagues from the build tools team arrive, and a discussion about espresso machines, barbecue grills or why we don't want to discuss Agile evolves. The day can come.

9:10 am: Time to do some work! And, no, that doesn't involve foosball. Or ping pong. I'm actually writing code. Unfortunately I am not allowed to chat about what we do, or even how we do it. So just imagine me typing. And talking to people. And typing again. More talking. More typing. You get the idea.

12:10 pm: The hardest question at lunch time is which cafe to choose. I just run with the crowd of people working next to me. Today we're heading over to Pintxo. One thing that constantly surprises me about food at Google is that I actually like the dessert. Chocolate cookies, sliced fruit, hot brownies, some chocolate cream topped with strawberries. Take your pick.

1:00 pm: Work, work, work. Getting some coffee. Work, work, work. More coffee. Work, work, work. Grabbing a snack, which leads to some discussion about pair programming and the state of the world in general. Work, work, work.

7:30 pm: Going to Charlie's to get my usual treat for the evening: a self-designed burger and a coke. This certainly feels like America. I meet Nicolai and his friend, who doesn't work for Google. I learn that we're allowed to bring visitors to dinner one or two times a month. We talk about real estate prices and I realize that living here is even more expensive than Munich. Dang!

8:15 pm: I finally arrive at the apartment. I kill half an hour by starting Yet Another Blog Post. I want to write something about unit test size that features some modestly comical adult references (go figure). After some time I realize that I don't have enough high energy content for this entry. I'll finish it eventually. Blogger's shortest joke.

9:00 pm: A new episode of House starts. For some strange reason the ability to watch a show that doesn't run in Germany yet makes me feel childishly happy. This probably reveals things about my personality I don't even want to think about.

10:00 pm: The favorite part of my day is talking to my wife. And this is PRIVATE! ...

10:30 pm: Another day at Google passed. I do some reading and eventually fall asleep. Just for the reference: my dreams are private, too. Just in case you'd hoped for something. Good night!

An Editor Independent Unittest Executor

Since I got test infected I'm somehow unable to write a single line of untested code without feeling uneasy. When I just want to write a tiny script containing a few lines of code in whatever text editor is installed in a system, it seems to be a daunting task to set up a programming environment that allows you to execute unit tests with a single click. But this single click is what makes writing unit tests unobtrusive enough to keep doing it.

So I'm quite fond of using a simple script to execute my script's unit tests whenever I save it. This concept is not new, and certainly not an original idea in itself, but the simplicity of an editor independent unit test executor in 10 lines of code has a certain appeal for me:

#!/bin/bash
stat_command="stat -c '%Y'"
file_name=$1
last_modification=""
while true; do
  current_modification=$( $stat_command $file_name )
  if [ "$current_modification" != "$last_modification" ]; then
    clear
    $file_name --test
    last_modification=$current_modification
  fi
  sleep 1
done

This script stats the script file until it detects a change. Whenever a change is detected, the script is called with --test, which is my personal way to tell a script that it should just execute it's unit tests and exit. See my blog post about integrating unit tests in Ruby scripts to learn how this can be done in Ruby. A very similar approach is possible for Python:

#!/usr/bin/python
import unittest
import sys

if sys.argv.count("--test") > 0:
  sys.argv.remove("--test")
  unittest.main()

Now I can simply call the test bash script, giving it the script under test as parameter:

./run_tests.sh ./script_under_test.py

The beauty lies in the simplicity of the solution: Even when I remote edit a script on some server with vi, I can simply launch a new console and execute run_tests.sh, watching the test results whenever I type ":w".

Update: The "sleep 1" really helps to keep I/O load down. Thanks to Philip for pointing this out. And yet another nice example of how hard it is to write 10 lines of bugfree code without a test.

Integrating Unit Tests In Ruby Scripts

When I write ruby scripts I like to use a single file, containing the program and all unit tests. It took me some time to find out how to add a command line switch to my ruby scripts that makes them run in script mode with full access to the Test::Unit command line arguments, while being able to run the script without the test framework interfering in the execution:

#!/usr/bin/ruby -w
require 'test/unit'

if ARGV.include?("--test")
  ARGV.delete_at(ARGV.index("--test"))
else
  Test::Unit::run = true
  puts "Running..."
end

Now you can simply run the program by typing

$ ./testme.rb
Running...

or run the tests with

$ ./testme.rb --test
Loaded suite ./testme
Started

Finished in 0.0 seconds.

0 tests, 0 assertions, 0 failures, 0 errors

The nice thing is that only the first "--test" will be removed, so you can still leverage the Test::Unit command line argument interface.

My First OCaml Tests - So Close To Heaven!

Some time ago a a discussion in the testdrivendevelopment Yahoo-Group evolved around the concept of "testable languages". I thought about this for a while and came up with the idea that I want to be able to have expressions as first class citizens:

assertThat { assertThat(false) } abortsWith TestFailedException

Today I played a little with OCaml, sorted my functional programming skillz out, and finally arrived at my first test driven unit test environment for OCaml! Note that it's nearly what I wanted to be able to write, but unfortunately there's not enough syntactic sugar for lambda expressions (or I didn't find out, yet), so I'm stuck with using the quite ugly ( function () -> expression ) syntax. But hey, it's really close to heaven.

exception TestFailed
exception TestError

let failsWith expectedError expression =
	try
		expression ();
		false
	with error ->
		expectedError = error

let isTrue expression: bool =
	expression

let isFalse expression: bool =
	not (expression)

let assertThat expression conditionMatchesOn =
	if not (conditionMatchesOn (expression)) then 
		raise TestFailed 
	else 
		()

let _ = (
	assertThat true isTrue;
	assertThat (isTrue true) isTrue;
	assertThat (not (isTrue false)) isTrue;
	
	assertThat false isFalse;
	assertThat (isFalse false) isTrue;
	
	assertThat (TestFailed = TestFailed) isTrue;
	assertThat (failsWith TestError (function () -> raise TestError)) isTrue;
	assertThat (failsWith TestFailed (function () -> raise TestError)) isFalse;
	assertThat
		( function () -> assertThat false isTrue )
		( failsWith TestFailed );

	Printf.printf "OK\n";
);;

XPDays Germany 2007 - Ideas Going Mainstream

I had the opportunity to take part in the XPDays Germany last week. The company I work for enabled Uwe, our project lead, Holger and me to participate. It all started with a three and a half hour ride from Munich to Karlsruhe where we heroically overcame a nearly empty tank, a shaking car that felt like it just drank the wrong kind of gas and my own card reading skills - or lack thereof.

Day 1

In the end we arrived on time for a Randori session by Dave Nicolette and Rod Coffin. From the moment I learned of this experimental learning session where two people sit in front of a computer and test drive a piece of code while the whole audience is throwing in questions, I was kind of scared of the prospect of being watched while writing code by a hundred people - which is probably kind of normal, given that some people even dislike being watched while pair programming.

The session turned out to be very interesting. One of the key elements of this style of learning is that the audience's energy level stays high for a long time - you have to pay close attention, since due to the random selection of the next person to come to the front you could always be this person. And since you don't want to look like a fool when doing stuff in front of a hundred people (um, what was the problem, again?) my adrenaline level alone was enough to keep me awake.

The other thing I learned from this experience besides a new way to coach technical stuff is that there is a very good reason we do "pair programming" and not "group programming". Throwing two brains at a problem can be a very mighty tool to solve programming tasks, but in some situations throwing a hundred brains onto a very simple programming example felt like sitting in one of Dilbert's most unproductive meetings:

"So we add the game to the character: character.addGame(game)"
"But why don't you add the character to the game?"
"I see duplication, I see duplication!"
"But isn't this, um, less expressive?"
"Duplication is bad!"
"Why do you assign null to a variable, this is done automatically"

After some time of silent observation I realized that sometimes I am exactly like this! So to all of you who have to cope with me on a daily basis: just hit me on the head with a big club from time to time.

The next session was a series of "lightning talks" about all kind of Agile topics where I learned about Alistair Cockburn's Crystal, which is a set of methodologies built upon the Agile principles and an extreme tailoring approach.

Day 2

Despite the enormous amount of a whole liter of "badisches Helles" I had in the evening I was wide awake and ready to suck in new ideas during the conference's main part. The day started with an interesting presentation by Dave Nicolette about how to communicate TDD and design debt to your management. I was particularly stunned by the fact that he did talk about the cost of design debt and the refactoring part in the TDD cycle for a very long time without mentioning the cost of fixing an error in terms of when it is found and the shortened feedback cycles that TDD provides.

The next presentation was titled "why Agile projects fail", but turned out to be about why projects fail in general and provided some insight into the ideas of root cause analysis (5 Why), the dimensions in which failure can occur (The Broken Triangle), and the psychological factors that are the real cause of ineffective development practices.

After lunch the keynote by "Dark Side" Rod Austin from HBS showed how Agile development fits the icy wind of change that swirls today's leading companies in the world from a cost competitive to an innovative business model. After all, who doesn't want a designer trash bin?

Stfan Roock's talk on "Simplicity in Software Projects" was a very entertaining lecture on how easy it is to get so accustomed to complexity that you don't even realize how simple things could be. Well, that and that the Borg are the only entities in the universe who understand that when you travel through space aerodynamics is pointless.

In the end it was very interesting to see big German companies like SAP, EADS and Siemens to take interest in extreme programming. Looks like those ideas are finally going mainstream.

Does "Test After" Work For You?

"Why not write a test for this?"
"Why should I, it works..."

The idea of Test-After Development is to write a set of automated white-box tests after writing your production code. Since probably every CS student in the world has learned that unit tests are a good idea, you'd expect unit testing to be an industry state standard for quite a while now. Interestingly the idea of automated unit and integration test is lately becoming more popular due to the widespread use of Test-Driven Development.

So why do we need Test-Driven Development to be able to efficiently write automated unit tests?

• If you write your code first and don't think about how to test the code, the code will not be testable. Thus testing becomes expensive and frustrating. Test-Driven Development will guide your software design by the old mantra of "how-do-I-want-to-use-this-class", leading to a highly decoupled design.


• When you write your tests, you'll discover a lot of errors. But instead of the red bar in Test-Driven Development, which you expect, the red bar in Test-After Development is the demotivating sword of reality.


• The most important reason why I have never seen Test-After Development work, is that developers just don't believe in errors once they wrote the code. This seems to be an eternal wisdom of software development psychology: once the code works, why bother testing it? Let's just implement the next feature.

A Program Is Born: QCMake's First Functional Test

When you write a GUI that is just a thin layer for an existing business layer and you don't see how to integrate test fixtures into this business layer, you'll be down in the dirty functional testing work very quickly. This happened to me today when I tried to write my first test for a small Qt facade object for cmake.

I started the test very enthusiastically: To test cmake I create a directory, cd into that directory, create a CMakeLists.txt and let cmake create a CMakeCache.txt. In the end I know that cmake ran when CMakeCache.txt exists.

void QCMakeControlTest::shouldExecuteCMakeInTheCurrentDirectory()
{
  QDir currentDirectory;
  QDir testDirectory(currentDirectory.path() + "/ExecuteInCurrentDirectory");
  QVERIFY(currentDirectory.mkdir(testDirectory.dirName()));
  QVERIFY(QDir::setCurrent(testDirectory.path()));
}

I hit F5 and everything runs just fine. Once. The second time the directory ExecuteInCurrentDirectory already exists. Of course to have a nice and clean starting point the test must remove the test directory if it already exists. So I added:

void QCMakeControlTest::shouldExecuteCMakeInTheCurrentDirectory()
{
  QDir currentDirectory;
  QDir testDirectory(currentDirectory.path() + "/ExecuteInCurrentDirectory");
  if(currentDirectory.exists(testDirectory.dirName())) 
  {
    QVERIFY(currentDirectory.rmdir(testDirectory.dirName()));
  }
  QVERIFY(currentDirectory.mkdir(testDirectory.dirName()));
  QVERIFY(QDir::setCurrent(testDirectory.path()));
}

Green. Perfect. Now let's create a CMakeLists.txt.

void QCMakeControlTest::shouldExecuteCMakeInTheCurrentDirectory()
{
  QDir currentDirectory;
  QDir testDirectory(currentDirectory.path() + "/ExecuteInCurrentDirectory");
  if(currentDirectory.exists(testDirectory.dirName())) 
  {
    QVERIFY(currentDirectory.rmdir(testDirectory.dirName()));
  }
  QVERIFY(QDir::setCurrent(testDirectory.path()));

  QFile cmakeLists("CMakeLists.txt");
  QVERIFY(cmakeLists.open(QIODevice::ReadWrite));
}

Green again. Once. The test fails the second time it's executed:

********* Start testing of QCMakeControlTest *********
Config: Using QTest library 4.3.2, Qt 4.3.2
PASS   : QCMakeControlTest::initTestCase()
FAIL!  : QCMakeControlTest::shouldExecuteCMakeInTheCurrentDirectory()
  'currentDirectory.rmdir(testDirectory.dirName())' returned FALSE. ()
..\..\..\..\..\Source\CMake\Source\QTDialog\qcmaketest\QCMakeControlTest.cpp(30) :
  failure location
PASS   : QCMakeControlTest::cleanupTestCase()
Totals: 2 passed, 1 failed, 0 skipped

Yep, no problem, all I need to do is to rmdir recursively. Just a quick glance into the Qt docs. But I found nothing. Well, it's not too hard to implement a recursive rm -rf, but still... I was so sure that this function would be hidden somewhere that I spent more time googling and doc-reading than implementing it when I finally realized that I was on my own. So in the end the test looked a little bloated:

#include "qcmaketest/QCMakeControlTest.h"

#include "qcmakeui/QCMakeControl.h"

bool removeRecursiveForced(QDir& directory, const QFileInfo& entry)
{
  if(!entry.isDir())
  {
    return directory.remove(entry.fileName());
  }
  QDir directoryEntry(entry.filePath());
  QList entries(directoryEntry.entryInfoList
    (QDir::Dirs | QDir::Files | QDir::NoDotAndDotDot));
  for(int entryIndex = 0; entryIndex < entries.count(); ++entryIndex)
  {
    if(!removeRecursiveForced(directoryEntry, entries.at(entryIndex)))
    {
      return false;
    }
  }
  return directory.rmdir(entry.fileName());
}

void QCMakeControlTest::shouldExecuteCMakeInTheCurrentDirectory()
{
  QDir currentDirectory;
  QDir testDirectory(currentDirectory.path() + "/ExecuteInCurrentDirectory");
  if(currentDirectory.exists(testDirectory.dirName())) 
  {
    QVERIFY(removeRecursiveForced(currentDirectory, 
      QFileInfo(currentDirectory, testDirectory.dirName())));
  }
  QVERIFY(currentDirectory.mkdir(testDirectory.dirName()));
  QVERIFY(QDir::setCurrent(testDirectory.path()));

  QFile cmakeLists("CMakeLists.txt");
  QVERIFY(cmakeLists.open(QIODevice::ReadWrite));

  QCMakeControl qCMakeControl;
  qCMakeControl.configure();

  QFile cmakeCache("CMakeCache.txt");
  QVERIFY(cmakeCache.exists());
}

#include "QCMakeControlTest.moc"

At least I have an idea where this could lead me - a nice class to generate a clean cmake directory. But let's see whether I'll be right, perhaps YAGNI will finally get back at me. And if you know an easier way to delete a directory recursively with Qt, please leave a comment.

Fasten Your Seat Belts! - Chances And Pitfalls Of Test Driving Your Development

Ten years ago I had a rendezvous with a beautiful girl, and at the end of the evening I gave her a ride home. Back than I thought playing the gentleman to be posh, so I opened the door for her. She slid with one elegant movement into her seat and I paced around the car and folded my wiry frame behind the steering wheel. I looked at her.

"Don't you want to start?", she asked and watched me curiously. "Um", I said, obviously always finding the right words at the right moment. "Um. - Not before you buckle up...". She frowned at me: "But I never buckle up". I replied "Well, if you don't buckle up, we're not gonna go anywhere tonight". "Oh come on!", she now somewhat furiously stated. "Nope!" I insisted eloquently, finally feeling her shield of stubborn resistance falter. It took a few seconds before she realized that I really wouldn't drive her home without her being properly protected from falling through the windshield and decomposing her pretty head by hitting the next best fireplug. So she buckled up.

Even back than I was so used to the secure feeling of the protective belt that just thinking of driving unbelted drove an uneasy quiver through my guts. This feeling is so strong that if I don't drive strapped when moving the car just a few inches there's always a sense of awareness that makes me want to fasten the seat belt immediately.

Today I felt exactly the same way while writing code.

The path of the test

At the beginning of the last iteration we identified a story that affected some legacy modules in our code base. When we recognized that the changes we needed to make would touch more code than we had thought, so we decided to try to test drive a part of the system from scratch to replace the tangled old code. So Richard, Reinhard and myself started to pair on the story alternately. Besides some private experiments with a Sudoku solver in Java this was the first time I was doing real full time TDD pair programming for a couple of days. Aside from some initial irritation and the constant realization that pair programming is hard to learn I was quickly pulled into the red-green-refactor cycle, as usual. But this time I held the pace for longer than ever. And was pulled into the cycle deeper and deeper. Write a test, make it work, look for redundancy. Write a test, make it work, look for redundancy. Write a test...

Today I wanted to quickly integrate the changed interface into an existing module. I didn't have a test yet. The cpp file was readily opened in my editor. Just a quick edit, nothing more than integrating this interface. A few simple edits. Only three lines or something. And suddenly a nagging question materializing in my head:

How can I make sure this works?

At this moment I felt like driving unbuckled. I felt unsafe. I wanted my cozy safety net back. Like an addict I went for the next test.

What use is a seat belt when you hit a tree at 200 MPH?

Since I'm the one driving the adoption of XP in our company, I wanted to try TDD for myself on a save playground to learn more about the ins and outs before applying it at work. Since Java has really nice tools for TDD, I started test driving a small Sudoku solver in Java. This was my first real test driven code and I often wondered about how nicely the test suite covered my errors. Spirited in the Agile fashion, I began with a really straight forward brute force implementation. Everything went a lot more smoothly than I had expected and after some coding I had a simple solution that needed over 90 seconds for one simple Sudoku.

After a while I wanted to optimize the runtime. So I introduced some caching variables. I struggled with the failing tests as my solution grew more and more sophisticated, but the tests helped me to get to a deeper understanding of the real problem. Finally I arrived at a point where the algorithm managed to work through 1400 Sudokus in less than a second. I was thrilled. And I wanted more. So I installed a profiling framework to find out where the next optimization sweet spot would be hidden. When I browsed the profiling data I realized that the real solver didn't even call the algorithm. So I had benchmarked a program that didn't solve any Sudoku at all.

At this moment I felt like hitting a tree with 200 MPH, suddenly realizing that it is not a good idea to drive that fast into a 90-degree turn on a wet street, even if you have a seat belt.

After the blood had returned to my head on it's way to my brain I implemented a test into the main program to check every solution with a simple algorithm before claiming to have solved anything. In the meantime I have a solution that runs 1400 Sudokus in 6 seconds on my core 2 notebook. I'm even quite convinced that I got the solution part correct...

Do I get my driver's license?

So, here's the lesson I learned on this journey on my path to the test:

• Don't rely on your tests too quickly.
  If you want to heed the XP advice to "test everything that can possibly break", be aware that it's often the things of which you think that they can't break that finally break.
  


• Use a healthy mixture of tests on all abstraction levels.
  Unit and functional tests are orthogonal - they cover different aspects of the code. But of course you'll already have a lot of unit and functional tests if you don't rely on your tests too quickly.
  


• Buckle up!
  The unsafe feeling while trying to modify code without having a test was a very impressive experience for me. I know that from now on I'll fasten my code's seat belt.
  

Can You Remember Hungarian TLAs?

Scene 1. Karl and J.B. are pairing on a small web service. Karl is just returning to the workplace with a steaming, hot cup of coffee in his hand.

Karl: 'Let me see what you wrote just now...

usUserName = request.getParameter("UserName");

Um... This variable, usUserName, what does the us-prefix stand for?'
J.B.: 'Well, that is the unescaped user name the way we get it from the user. I wanted to make sure that we don't accidentally write it into a database or send it back in it's evil, unescaped form to the webbrowser. If we use the us-prefix every time we have an unsafe string, we'll immediately recognize any error that could otherwise escape us because we will learn to look for such errors. This is the application Hungarian notation I read about over at Joel's site, where you actually use a prefix that has a meaning instead of just a shorthand for the type.'
Karl: 'So... why not just call it unescapedUserName?'

Confusion

In a time where you enter veryLongVariableNames by typing 'v', 'e', 'r', Crtl-Space, I don't see why we can't finally get rid of TLAs. You know that the Hungarian notation got out of hand when your colleagues check in code that changes ucpBuffer to pucBuffer ("fixed a segfault"). Why not just name a variable for what it contains, in plain old English? I definitely know that I should think about my method name if my partner asks during a pairing session: "And what exactly do you intend to do in this method?".

In which example is the error easier to spot? Does the second example really take longer to write? To read? To understand?

for(unsigned int i = 0; i < iLineCount(); ++i)
{
  for(unsigned int j = 0; j < iNodeCount(i); ++i)
  {
    pGetNode(i, j)->layout();
  }
}

for(unsigned int lineIndex = 0;lineIndex < getLineCount(); ++lineIndex)
{
  for(unsigned int nodeIndex = 0; nodeIndex < getNodeCount(lineIndex); ++lineIndex)
  {
    getNode(lineIndex, nodeIndex)->layout();
  }
}

If you can really remember mnemonic prefix TLAs (or any TLAs for that matter) and think Hungarian notation or abbrVarNames are a great way to safe yourself some typing, please let me know.

Struggling to TDD a GUI application

CMake is one of the best build tools out there. It has a nice command line interface and comes with an even nicer GUI. Unfortunately the GUI is MFC based, which means you need a VC professional license to build it for windows and you can't use it in linux.

Since Trolltech released it's wonderful GUI framework Qt for windows open source development some time ago, I decided to combine my eagerness to learn TDDing GUI apps with my need for a nice cmake GUI - and to start developing qcmake.

The first priority for me was to learn how to TDD a GUI application. CMakeSetup, the MFC application qcmake should be able to replace, has a very simple single-window interface, so this should be the ideal playground to get an idea of the basic GUI testing problems.

Setting up the testing framework.

The first step to successful TDD is to set up a test environment where you can execute your tests with a single keystroke from within your development environment. I spent some time integrating Qt's testing framework qtestlib into ctest. Hitting F5 from my Visual Studio Express executes all the tests. If something goes wrong, the qtestlib framework prints the debug output into the Visual Studio output window. This way I can just click on the error message to find the offending code, or just enable a breakpoint step through my personal mess...

Top-Down or Bottom-Up - the duck's decision

The testing framework is ready and eagerly waiting for it's first real test. But somehow I don't know where to start. The options are quite simple: either the good ol' bottom-up approach, implementing one layer upon each other until I reach the top, or the top-down development 2.0 methodology where everything is faked or mocked, slicing the whole vegetable vertically until the feature is finished.

Since the top-down approach resembles the design-driven process the most (plus the running tests, minus some heavy documents) and Heusser & McMillan's presentation Interaction Based Testing at GTAC made my mouth water (I really like chocolate flakes), I thought I'd go for the top-down method.

My first user interface test

And finally my first test looks like this:

#include "QCMakeTest.h"
#include "QCMakeWidget.h"

#include <QtTest/QTestMouseEvent>

void QCMakeTest::shouldEmitConfigureSignalOnConfigurePressed()
{
  QCMakeUi::QCMakeWidget* qCMake = new QCMakeUi::QCMakeWidget();
  QSignalSpy configurePressed(qCMake, SIGNAL(configure()));
  QTest::mousePress(qCMake->getConfigureButton(), Qt::LeftButton);
  QCOMPARE(configurePressed.count(), 1);
}

QTEST_MAIN(QCMakeTest)
#include "QCMakeTest.moc"

That was a lot of work just to get started with a simple test and basically no functionality. Fortunately I have some TDD experience to build upon, and right now this experience tells me that the up-front effort will pay of in the short run due to not debugging a lot. Up-front effort, quicker development, isn't that what BDUF was all about? I'm curious where all this will lead me to...

Big Design Up Front vs. Just Enough Design Initially

Software is complicated. More often than not it's a complicated mess. Sometimes even a tangled complicated mess. And wherever you look all you see is tradeoffs. There are no easy solutions (tm). The dll hell is replaced by the side-by-side hell. Emacs is better than Vim, Vim is better than the Visual Studio editor and the Visual Studio editor is better than Emacs. The Visual Studio editor even has a kill-ring (Ctrl-Shift-Ins). But Emacs has a web-browser. Vim is way cooler because I can't remember the commands, even though they're orthogonal. To what? Why not write a new editor in Erlang. Well, no, not me, I just want an editor that has all the features of Emacs, Vim and Visual Studio. Now. But without the bloat of Emacs. Or Visual Studio. More slick, just like Vi.

Since the early days of computer science, when software developers still had to wear suits at work and wrote A.I.s in Cobol, um, COBOL, with their feet, people tried to find out how software development could be made less complicated. And they soon discovered that the secret sauce is abstraction.

Abstraction

Layers. Components. Modules. Interfaces. Design. Architecture. It's so easy: define an architecture, think of layers, interfaces, modules. Create a nice design that meets this architecture's goals. Hire a bunch of developers to implement the components.

From this level of abstraction it really sounds easy. This is why it's called abstraction: it hides the complicated details. The good thing is that as long as you work on this level of abstraction, it's cheap to change your concept. Or as Joel Spolsky says:

Designing a feature by writing a thoughtful spec takes about 1/10th as much time as writing the code for that feature—or less.

Well, than it's obviously a very good idea to do all the design first. After all, if you change your design, you'll have to change your implementation. As long as you didn't start writing code, changing your design is easy. Or even better, start at the architecture level. Hire the best consultants to create the perfect architecture. Hire some really bright guys to do your design. In the end, a bunch of monkeys can do the implementation. The dream of the pointy-haired boss came true!

"Um. Sounds easy. So, how do we know that our design is good?"
"This is easy: experience."
"But to get experience I'd have to actually try the design, won't I?"
"Yes, of course."

"So, if my design is not perfect in the first place, I'll learn this only when I try to implement it?"

"Well, yes, come to the point."

"Then how can I finish my design before the implementation phase?"

"Um. Well. You just do iterations. Big iterations, I guess, because design is so much easier to change."

"So I work for months on a design of which I don't even know that I will be able to implement it?"

"Perhaps... easier to. Um, change..."
"And when I finally find out that my design was crap, I'm in the implementation phase, a deadline looming on the horizon and no time to change the design and all the code that was already written?"
"... - well, is there a different way?"

Feedback

Tradeoffs again. Working with abstractions means to get less feedback. "I'll take the chair and hit the sentinel" will be a hard job if the chair turns out to weight a hundred pounds.

And feedback is important. One of the laws of software development is:
The longer it takes until you find out that you made an error, the more costly it is to fix that error.

This means that you should try to find your errors as quickly as possible. But when you're working on a high abstraction level, you just don't know all the complicated details because, well, that's why you're working on that high abstraction level, isn't it? So you'll find out that your design is crap when you're in the "implementation phase", at which point nobody has time to change the design. So you just live with the crappy design and run around cursing the designer and hating your job.

Fail!

One solution to discover your errors early, is to do Ultra Extreme Elite Programming (Joel Spolsky). Design just enough up front that you get an idea of where you're going, write the target down as a test and sit down with a colleague to find a redundancy-free implementation. When you find out that your initial design is crap, which you'll do very quickly, rely on your tests to help you refactor your code to a better design. Of course, as Joel puts it so beautifully, this is like driving around with the handbrakes on.

The question is whether driving around with the handbrakes on is really slower than driving at full speed with closed eyes and a plan. I think it mostly depends on where you want to end.

Defects On Sale!

Today after our planning game I did a short poll on how the guys perceive test driven development and pair programming. We're trying to do both for some time now, and since I take the blame for introducing both practices, I feel I'm somewhat - um - preoccupied on that matter. A few days ago, I was caught totally off guard when Richard told me that, well, he doesn't believe programming in pairs is more productive. Bummer. And I had believed my show to be grand circus.

Understanding The Fuzz About Engineering In Software Development

Steve McConnell responds to Eric Wise's article Rejecting Software Engineering that Rumors of Software Engineering's Death are Greatly Exaggerated". There's a lot of fuzz about the usage of the word "engineering" when it comes to software development. What's this all about?

What is engineering?

According to wikipedia, engineering is defined by the ECPD as

The creative application of scientific principles to design or develop structures, machines, apparatus, or manufacturing processes, or works utilizing them singly or in combination; or to construct or operate the same with full cognizance of their design; or to forecast their behavior under specific operating conditions; all as respects an intended function, economics of operation and safety to life and property.

When I look at this definition I don't really see anything that would not be applicable to software development. Steve McConnell generously points out that software engineering is recognized and practiced for some time now, so where does this new-fashioned stubborn refusal to call the child by it's name come from?

The engineering process

The real problem comes to light when you look at the engineering process. You'll find a description of the engineering process on wikipedia. The article describes the engineering process in four stages:

• Conceive


• Design


• Realize


• Service

Now there are some engineers who map the engineering stages to software development in a rather funny way: They think that "Design" is the process of drawing good looking UML diagrams and that "Realize" is "Coding". When you look at the wikipedia article, you'll see that for engineers "Realize" stands for "Manufacturing". In a software context, manufacturing means running the compiler and pressing some CDs or deploying some binary over the Internet.

So when engineers claim that software developers should look at how engineers do their design and all this talk about software processes would be settled once and for all, they're ignoring that software development is a design-only activity and that software has a lot less problems with the "Realization" stage than traditional engineering.

When a software developer writes code she is building an executable, mathematical model of reality.

Conclusion

When software developers prefer not to use the title "engineer" to describe what they're doing, they're trying to avoid a mapping of the engineering process onto the software development process that is wrong. In the end, software developers will build mathematical models (source code) and apply scientific methods (complexity analysis) to solve problems. If this is not engineering, than we're not engineers.

How You Can Start Improving Your Software Process Today

Are you a developer who dreams about a better software development process in the organization you work for? Maybe you read something about fancy practices on your favorite blog or mayhap you even touched one of those old-style paper collections called books? Do you have some concrete ideas on how to improve, but don't know how to start? I was in the same situation a year ago. Here's what I did and what I would do differently today.

Today The Test Suite Broke

When I arrived at work today I fired up Outlook and checked my mail. I found five mails from our auto-build server telling me that the build broke. Since we introduced test driven development and continuous integration only a short time ago this was not out of nowhere - the build usually breaks at least once a day.

But today was the first day a unit test broke since we introduced TDD and CI.

Ubuntu Gutsy Ximeta NDAS Howto

A month ago I bought a TREKSTOR NDAS device. This devices promises on it's package to be linux compatible. So after I unpacked the hardware and everything was running in Windows I tried to install it in linux. Unfortunately the stock feisty debian package I found didn't work with my WLAN configuration.

Now after I reinstalled ubuntu and upgraded to gutsy which comes right now with a 2.6.22er kernel I tried to build the driver from source. I had to patch the sources to make it work, but since it works flawlessly right now I provide my patch and a little compilation howto.

Download the current NDAS sources and my NDAS patch for linux kernel 2.6.22.

# installed some packages. I don't know which exactly, but you'll need
# at least the following:
apt-get install build-essential checkinstall linux-headers-generic

# extract and patch the ndas sources...
tar xvzf /path/to/ndas-1.1-2.tar.gz
cd ndas-1.1-2
patch -p1 < /path/to/ndas-1.1-2_kernel-2.22.patch

# you only need to set NDAS_KERNEL_VERSION if you 
# don't want to compile ndas for the currently running 
# kernel for example, if you're compling from within colinux
NDAS_KERNEL_VERSION=2.6.22-6-generic
make

# ndas_root must be exported for make install and 
# checkinstall to work
export ndas_root=$(pwd)
# somehow I had to make install before checkinstall...
# this is no problem, since checkinstall will clean up
# the whole mess again
sudo make install
sudo checkinstall

After that you can start the NDAS service by issuing

/etc/init.d/ndas start

Configure your device by following the Ximeta NDAS driver documentation.