curious-attempt-bunny

Gradle Goodness

2012-01-20T09:02:00.001-08:00

This is my final installment, and it's a biggie! This post includes:
* automatically opening test reports on test failure
* grouping tasks for the gradle task list
* renaming all your build.gradle files to match the names of the subprojects
* plugin extensions as the replacement for plugin conventions
* changing the project's artifactId (useful for non-uploadArchive tasks consuming the pom)
* dependency specifications for easy reuse of customizations such as exclusions in multi-project builds

Automatically opening test reports on test failure. As an added bonus, your headless CI server will return false for Desktop.isDesktopSupported(). Note that this requires JDK 1.6:

test {
    successful = true

    afterTest { desc, result ->
        if (result.toString() == 'FAILURE') {
         successful = false
        }
    }
}

import java.awt.Desktop

gradle.taskGraph.afterTask { task, taskState ->
    if (task instanceof Test && !task.successful) {
        if (Desktop.isDesktopSupported()) {
            Desktop.desktop.browse( new File(task.testReportDir, 'index.html').toURI() )
        }
    }
}

Grouping tasks for the gradle task list. Make use of the Task.group property:

task build(dependsOn: [test, assemble]) {
    group = 'build'
}

Renaming all your build.gradle files to match the names of the subprojects. Place this in your settings.gradle:

rootProject.children.each { project ->
    String fileBaseName = project.name.replaceAll("\\p{Upper}") { "-${it.toLowerCase()}" }
    project.buildFileName = "${fileBaseName}.gradle"
}

Plugin extensions as the replacement for plugin conventions:

project.extensions.myextension = [name: "default"]

myextension.name = "some other value"

myextension {
    name = "a different value"
}

Changing the project's artifactId (useful for non-uploadArchive tasks consuming the pom):

project.archivesBaseName = "my-preferred-artifactId"

Dependency specifications for easy reuse of customizations such as exclusions in multi-project builds. Use: dependencies.create(notation, closure).

Enjoy!

Gradle Goodness

2012-01-18T18:13:00.000-08:00

A few more of today's learnings in Gradle:

When migrating build script code to a plugin you will typically need to add project. in places (because the delegate of a Gradle build script is the project). You can use Project.configure to make the migration smoother, e.g.:

apply plugin: MyPlugin

class MyPlugin implements Plugin {
  void apply(Project project) {
    project.configure(project) {
       version '1.0' // instead of project.version '1.0'
    }
  } 
}

For testing your Gradle plugins you can use ProjectBuilder.builder().build() to test the wiring of your plugin. For integration testing there's GradleDistributionExecutor, BUT it's not publically exposed - so please vote for the testing toolkit roadmap item to see this feature sooner.

So you know about the Copy task. Ever find yourself wanting to be certain that the destination folder is clean of other output? use the Sync task. Very nice.

Find yourself wanting to configure all current and future tasks matching a certain criteria? Take a look at the project.tasks. E.g.:

tasks.withType(Jar) {
    aProperty = 'value'
}

Enjoy!

Gradle Goodness

2012-01-17T20:12:00.001-08:00

A little known fact is that in Gradle you can now change property assignments into method calls in your build scripts (essentially by omitting the equals sign). Here's the good part: if the property does not exist the build script will fail. Yay! No more silent typo bugs! To illustrate, this works:

version = '1.0'

As does this:

version '1.0'

But this build script will fail:

versionxxx '1.0'

Nice!

Another tip:

Rather than make use of afterEvaluate you may be able to make good use of lazy GStrings. For example you could define your project version as "version-${-> buildNumber}"! See the explanation of Groovy lazy GString syntax for more information.

Did you know that the --exclude-task (-x) command-line switch differs from skipping tasks in that it prunes away the whole task graph associated with the excluded task? I didn't!

Finally, if your company is putting a lot of effort into customizing the Gradle experience, did you know that you could take advantage of Gradle init scripts in the Gradle init.d folder? Yes? Fiddly though, isn't it? Did you know that you can bundle your custom init scripts inside a Gradle distribution?!

To use this you need to assemble a Gradle distribution zip (basically by adding your init.d scripts to it), make it available internally (say, in your Artifactory repository), and then simply edit your gradle/wrapper/gradle-wrapper.properties. In there is the URL to download the Gradle distribution (distributionUrl).

Enjoy!

Hackergarten in PDX - Friday January 20th

2012-01-12T19:52:00.000-08:00

I'm taking advantage of Luke Daley coming to town next week, and organizing the first US Hackergarten here in Portland. The idea is to come up with an improvement to an open source project in a short period of time. I plan to work on GRADLE-1677. So far participants include:
* Luke Daley (of Geb, and Gradle)
* Howard Lewis Ship (of Tapestry)

YesMail is hosting, and we kick off at 6pm on Friday January the 20th. Are you interested? RSVP and I'll send you all the details.

Machine Learning: Applying Anomaly Detection to Real World Problems

2012-01-09T17:37:00.000-08:00

I've had a lot of success applying an Anomaly Detection algorithm to a problem at work. Without revealing anything sensitive, essentially the idea is be able to flag actions taken by the user of the system as "anomalies" and take preventative action.

There are many ways to do anomaly detection. My data has some characteristics that made the Multivariate Gaussian approach a good fit, namely:

* most features follow a normal distribution
* anomalies have features with values on extreme ends of the scale (i.e a number of standard deviations away from the mean)
* feature values are interrelated (i.e. a large value in feature A tends to correspond to a large value in feature B)

I didn't apply Principle Component Analysis. I wanted to see how good the results were before complicating the solution. As it turns out, it does pretty well. I had 4271 examples - of which only 3 were labelled as known anomalies. Only 0.5% (24) examples were predicted to be anomalies. All three known anomalies were found, and a further four were uncovered after hand labeling the remaining 21. Very nice indeed!

This results in a fantastic seeming F1 score (0.998). The fallacy in the calculation is that I haven't hand labeled a representative set of the unlabeled data - i.e. I've made the dangerous assumption that all examples predicted to be normal are indeed normal, and that there were no anomalies in fact predicted to be normal.

Very rewarding, none the less. All of this enabled by Octave of course. Ahh: hist(X(:,2))

Node: Reading Every Line in a File

2012-01-09T12:18:00.000-08:00

In Groovy this would be concise: new File(filename).eachLine { ... }. Here's what it looks like in Node using CoffeeScript:

fs = require('fs')
EM = require('events').EventEmitter
ev = new EM()
stream = fs.createReadStream(filename)
buffer = new Buffer(0)
stream.on 'data', (data) ->
    nextBuffer = new Buffer(buffer.length+data.length)
    buffer.copy(nextBuffer)
    data.copy(nextBuffer, buffer.length)
    start = 0
    offset = buffer.length
    buffer = nextBuffer

    for i in [1..data.length]
        if data[i] == 10
            end = i+offset+1
            line = buffer.slice start, end
            start = end
            ev.emit "line", line.toString()

    buffer = buffer.slice start
            
stream.on 'end', () ->
    ev.emit 'line', buffer.toString() unless buffer.length == 0
    
ev.on 'line', (line) ->
    ...

Do you have a smarter way of doing this?

Edit: I'm reading a very large file with this code. That's why this isn't synchronous.

Escaped HTML in Jade Layout Body

2012-01-03T12:38:00.000-08:00

Jade makes for very readable HTML. The one hurdle that got me was how to reference the body of your view without escaping the HTML. I had been using #{body}, whereas this does the right thing: div.postings!= body.

CoffeeScript Frustrations (in TextMate)

2012-01-03T11:57:00.000-08:00

If you're using CoffeeScript in TextMate and you're getting some wierd behaviour (for me are variables being undefined), then TextMate's tabbing may be the problem. First, verify that this is your problem by inspecting the generated JavaScript. If it's the tabbing, switch TextMate to soft tabs.

Heroku: Parsing RSS Feed XML

2011-12-26T14:34:00.000-08:00

I want a scheduled job to run on heroku that pulls in data from an external RSS feed. I started out with libxmljs. Things looked promising until I deployed to heroku. In order to build my slug file libxmljs needs to have access to SCons:

-----> Installing dependencies with npm 1.0.94
       
       > libxmljs@0.4.3 preinstall /tmp/build_16f6d4f95rt18/node_modules/easyrss/node_modules/libxmljs
       > make node
       
       make: scons: Command not found
       make: *** [node] Error 127
       npm ERR! error installing libxmljs@0.4.3 Error: libxmljs@0.4.3 preinstall: `make node`
       npm ERR! error installing libxmljs@0.4.3 `sh "-c" "make node"` failed with 2

Shame!

Easyrss looked good too but also had the same libxmljs dependency.

In the end I settled on xml2js which works just fine. Here's my code snippet:

var request = require('request'),
    xml2js = require('xml2js'),
    _ = require('underscore');

var parser = new xml2js.Parser();
request('http://url.to.my.rss.feed', function(error, response, body) {
    if (!error && response.statusCode == 200) {
        parser.parseString(body, function (err, result) {
            _(result.item).each( function(item) {
                console.log(item.title);
            });
        });
    }
});

I can exercise the script successfully via heroku run bash.

Learn You a Gradle Fu For Great Good

2011-12-21T08:52:00.000-08:00

It's not too late to sign up for the Gradle training course that's coming to PDX in January*! The training came up at last night's PJUG (although Gradle builds for JVM languages in general). Here is my on-list response:

"That [500 line] pom.xml from last night is a great reminder of what is was like to use Maven. Maven is great at dependency management, but terrible at everything else. Gradle build scripts are readable, concise, only go into detail where it matters, and are configurable at a higher level of abstraction than Ant. Perhaps most importantly, Gradle is well documented.

All that said, there are conventions, idioms, and a certain way of thinking that will heighten your Gradle-fu. You will learn these on your own over time. However, attending the training course is going to be much more time efficient."

If you don't know what this is all about, then you should definitely be taking a look at Gradle.

* Howard Lewis Ship recommends Gradle too.

1. Heroku + Gradle + Graffiti + AI 2. ? 3. Profit!

2011-12-20T21:51:00.000-08:00

Very nice. James Ward came to evangelize Heroku at our local usergroup. Thanks, James! Armed with a better understanding of Heroku, I set out to provide a simple web-service on heroku while at the usergroup post-talk bar visit. So, here it is:

GenderService! Okay. It's a dubious name, but it does something useful. You can hit the webservice and provide a name and it will infer the gender for you. You can make up a name that doesn't even exist and it will do a reasonable job of determining the gender. Give it a shot!

Here are some that I tried: (if you browser behaves strangely, it may be detecting that it's JSON and trying to be helpful)

In case you're curious, I'm using ngramms, laplace smoothing, and a smallish labelled dataset to estimate male/female probabilities for a name. If that's interesting to you, and you're local to PDX, consider joining pdxai on googlegroups.

Edit: I've added a confidence to the response. Clearly it's certain that I'm female: {"name":"merlyn","male":false,"confidence":0.995}

Building a Self-Driving Remote-Control Car

2011-12-19T07:59:00.000-08:00

Impressive! Take a look at the full blog post.

Learn Me a Haskell

2011-11-25T20:38:00.000-08:00

* f a is like f(a) in Java/C,
*

map f [a, b] is the same as [f a, f b],
* a $ b $ c the same as a (b c),
* (f . g) x is the same as f (g x),
* digitToInt '1' == 1,
* "hi" == ['h', 'i'], and
* show 100 == "100".

So, all these expressions evaluate to 6:Prelude Data.Char> all (==6) [
    sum (map digitToInt (show 123)),
    sum $ map digitToInt $ show 123,
    sum . map digitToInt $ show 123,
    (sum . map digitToInt . show) 123]
True
It's surprising is how simple the definition for $ is:
f $ x = f x

This Is The Life..
2011-11-15T14:07:00.000-08:00

Sure it's a little chilly this time of year, but coding out on the deck by the fire. Nice!...

Cheating at the Prime Directive?
2011-11-02T21:16:00.000-07:00
So it turns out that an analytic solution to the ant fighting for aichallenge.org is, well, non-trivial. My new ant bot now does a pretty reasonable job of it, yet the lazy part of me wonders if there's an easier way.

I've been taking the online Machine Learning course at Stanford, so naturally I wanted to apply last week's theory to it. I wouldn't expect a linear classifier (a one-vs-all linear logistic regression classifier) to actual perform all that well. But it does.

So. I took replays JSON of games played by the #1 ranked bot. I put the replays through a simple-ish Groovy script, and every time an ant was in proximity to an enemy ant I captured the local surroundings and the corresponding order as a training data example. A few games yielded over 100k unique training data points - which I shuffled to remove any bias in the distribution for the training and test data sets.

Using Octave the actual programming for the training was very very simple. Here's a taste of how simple (I haven't posted the sigmoid.m, costFunctionReg.m, or predict.m because they're coursework answers):clear ; close all; clc

data = load('training.all.nodupes.shuffled.txt');
yRaw = data(:, [1]); X = [ones(size(yRaw)) data(:, 2:size(data,2))];

trainSize = 110000;
testSize = length(yRaw)-trainSize;

lambda = 300;

fprintf('Training set size: %d\n', trainSize);
fprintf('Test set size: %d\n', testSize);
fprintf('Regularization lambda: %d\n', lambda);

for yTarget=1:5
  fprintf('Target classification label = %d ...\n', yTarget);
  y = yRaw;
  for i=1:length(y)
    if (y(i) == yTarget)
      y(i) = 1;
    else
      y(i) = 0;
    end
  end

  trainX = X(1:trainSize, :);
  trainY = y(1:trainSize, :);
  testX = X(trainSize+1:trainSize+testSize, :);
  testY = y(trainSize+1:trainSize+testSize, :);

  [m, n] = size(trainX);
  initial_theta = zeros(n, 1);

  options = optimset('GradObj', 'on', 'MaxIter', 1000);
  [theta, cost] = ...
      fminunc(@(t)(costFunctionReg(t, trainX, trainY, lambda)),
          initial_theta, options);

  fprintf('Train set accuracy: %f\n',
      mean(double(predict(theta, trainX) == trainY)) * 100);
  fprintf('Test set accuracy: %f\n',
      mean(double(predict(theta, testX) == testY)) * 100);

endAnd this is the output: (the target classification labels here are N, E, S, W, and - respectively)octave-3.4.0:36> train
Training set size: 110000
Test set size: 27180
Regularization lambda: 300
Target classification label = 1 ...
Train set accuracy: 79.710909
Test set accuracy: 79.337748
Target classification label = 2 ...
Train set accuracy: 78.755455
Test set accuracy: 78.491538
Target classification label = 3 ...
Train set accuracy: 80.984545
Test set accuracy: 81.111111
Target classification label = 4 ...
Train set accuracy: 78.599091
Test set accuracy: 78.863135
Target classification label = 5 ...
Train set accuracy: 82.936364
Test set accuracy: 83.153054Impressive! Well, the proof is in the pudding. Once this is integrated into an Ant bot I first need to see if it can beat my current Bot.

If you're reading the Octave source code and feeling cheated, yes, the magic happens inside the fminunc function. This is part of the beauty of Octave. You don't have to sweat to write a hyper-efficient numerical methods implementation. You don't have to wade through questionable third-party libraries for something that might work. You can use one that's provided and does the job well. Very nice indeed.

Ignore the Prime Directive!?
2011-10-29T07:32:00.000-07:00
My bot on aichallenge.org is doing a reasonable job (rank 205 of 3558 and climbing). Yay! Interestingly, it's doing so well despite not trying to score points and being totally unaware of the enemy.

Here's a recent replay (my bot is DaBunny, in white):

Edit: One day later, no changes to the bot code and it's now climbed to rank 93! It's on the main rankings page. It's rank climbs in 9 out of 10 games it plays. Howsat!?

Edit: Some weeks later, the most recent incarnation of my bot is in the top 1% (45th of 6000+) and continues to climb (although not much further, I bet):

Octave on Mac OSX: Aquaterm window opens but GnuPlot window is not displayed
2011-10-24T06:34:00.001-07:00
This one-liner fixed the problem for me:sudo ln -s /Applications/GnuPlot.app/Contents/Resources/bin/gnuplot /usr/bin/gnuplotThanks to Ben Sheldon for the solution.

NodeJS: Heroku push rejected, no Rails or Rack app detected
2011-10-10T21:33:00.001-07:00
If you're seeing this error when you git push heroku master:-----> Heroku receiving push
 !     Heroku push rejected, no Rails or Rack app detected... or this error listed by heroku logs:

heroku[slugc]: Slug compilation failed: no Rails or Rack app detected

Then most likely you didn't explicitly specify Cedar as your heroku stack. You can fix this now by doing the following:

$ heroku create --stack cedar
Creating freezing-water-1234... done, stack is cedar
http://freezing-water-1234.herokuapp.com/ | git@heroku.com:freezing-water-1234.git

Now you need to correct your git remote definition for heroku. I just edited my .git/config file, like so:

...
[remote "heroku"]
 url = git@heroku.com:freezing-water-1234.git
...

Now git push heroku master gets you on the right track.

Hope that helps!

Kill Comedy
2011-10-10T20:33:00.000-07:00
Seriously!?!

~$ ps -ef | grep sudo
    0 33853     1   0   0:00.01 ??         0:00.01 sudo gem install heroku
    0 33880     1   0   0:00.01 ??         0:00.01 sudo kill -9 33853

Perhaps this time?

~$ sudo kill -9 33853 33880
^C^C^C^C^C^C^Z

Grr...

Online Stanford AI Course Very Promising!
2011-10-10T20:09:00.001-07:00
The online Stanford AI course started today, and it's looking very promising! The format works well and is particularly effective. They make good use of diagrams, and I particularly liked the chinese translation example with the soup menu. Keep up the good (free!) work, Stanform.

Gradle: One-Time Maven2 pom.xml Dependencies Export
2011-10-06T09:05:00.000-07:00
Got Maven? Want Gradle? If so, the first bump in the road is extracting all your pom.xml dependencies for your build.gradle. The maven2gradle tool aims to do this (and more) for you, however if you have a Maven2 pom.xml then you may be out of luck for the time being.

In the interim this Groovy 1.8.x script works fine for simple pom.xml files:def project = new XmlSlurper().parse('pom.xml' as File)
def dependencies = project.dependencies.dependency
def compile = dependencies.findAll { it.scope?.text() != 'test' }
def testCompile = dependencies.findAll { it.scope?.text() == 'test' }

def printDependency = { type, dep ->
  def identifier =
     "'${dep.groupId.text()}:${dep.artifactId.text()}:${dep.version.text()}'"
  if (!dep.exclusions.size()) {
    println "  $type $identifier"
  } else {
    println "  $type($identifier) {"
    dep.exclusions.each {
      println "    exclude module:'${dep.groupId.text()}', " +
        "artifact: '${dep.artifactId.text()}'"
    }
    println "  }"
  }
}

println "dependencies {"
compile.each( printDependency.curry('compile') )
println()
testCompile.each( printDependency.curry('testCompile') )
println "}"

NodeJS: Error: Cannot find module 'utils'
2011-10-02T10:02:00.001-07:00
I get this error following the examples in Node Up And Running:Error: Cannot find module 'utils'
I presume utils was renamed to util at some point, because this works for me (in v0.4.12):var utils = require('util');

Gradle Build Script Style
2011-09-20T21:20:00.000-07:00
My thinking follows these lines: I put at the top the parts that will easily help the reader classify what kind of project it is. So, I put plugin declarations like groovy, java, war, etc at the top.

The next most likely thing I think the reader wants to know when they open a build script is what the dependencies of the project look like. That's why I keep my repositories and dependency definitions very close to the top of the script. I group compile and testCompile dependencies separately, and within those I promote subproject dependencies to the top. I haven't settled on a consistent style for where I place dependencies that I use for JavaExec or ant.taskdef.

The rest I consider to be the details and the reader can scroll down for those. I keep them following a logical flow from compile and processResources through to assembling artifacts.

After those come the details of artifact publishing and IDE configuration. These can grow to be big chunks of configuration, that IMO aren't too interesting to the reader. In that case I pull them out into separate build scripts and include them using apply-from.

The very bottom of the script is where I place functions I've defined and version and task class implementations that are small enough to live in the build script.

I avoid documenting anything that someone familiar with Gradle would know. I only comment work-arounds or awkward fragments of configuration - anything that someone would miss the significance of in a dangerous way, or anything overly cryptic. A single-line comment normally suffices. Task descriptions I view as optional if the task name itself is explanatory. Task descriptions are a must for rules - however simple they may seem to the author.

Gradle: A POM cannot have multiple artifacts with the same type and classifier.
2011-09-20T21:14:00.000-07:00
If you're getting this error then likely you were expecting to be publishing your custom jar or zip distribution:
A POM cannot have multiple artifacts with the same type and classifier.You can go the route of adding a pom filter to promote your artifact and implicitly suppress the publication of the java plugin's "jar" artifact. Yuck.

Alternatively you can remove the jar artifact from the archives configuration. Due to the current Gradle API this is the most concise I've gotten it:configurations.archives.removeArtifact( 
  configurations.archives.allArtifacts.find {
    it.file == tasks.jar.archivePath } )

Running Gradle in a Debugger
2011-09-20T20:25:00.000-07:00
Launch org.gradle.debug.GradleRunConfiguration from the Gradle project, pointing it at your build script using -b PATH_TO_SCRIPT/build.gradle.

Three Git "Tricks" I Can't Do Without
2011-09-19T20:05:00.000-07:00
For various reasons, I prefer command-line access to Git. The main "tricks" I use are (in order of fantasticness):
I make heavy use of git hist and git hist --all. For this you need to add an alias to your .gitconfig:[alias]
 hist = log --pretty=format:\"%h %ad | %s%d [%an]\" --graph --date=short

I help avoid mistakenly working on the wrong branch by including the branch name in my terminal prompt. I add this to my .bashrc or .profile file:function parse_git_branch {
 git branch 2> /dev/null | sed -e '/^[^*]/d' -e 's/* \(.*\)/(\1)/'
}

PS1="\w \$(parse_git_branch)\$ "

When I want to look at a file on another branch or in a particular commit, I use git show <branch/commit>:<full path>. For example:
$ git show master:build.gradle
Close contenders for the top three are git clean -n and git cherry-pick <hash>.

AI Challenge
2011-08-21T14:14:00.000-07:00
There's a simple dice game that you could try your hand at coding an AI for. Interested?

The game is Zombie Dice. You keep rolling until you either cash out with all your collected brains or you come up with three shotguns and get zero points.

Improve on my base-line actor!def actor(state) {

  if (state.shotguns == 2) "stop"

}The simulator is here on GitHub.

It could be that this problem is too simple. This above benchmarks at an average score per turn of 2.4. This benchmarks at 2.5:def actor = { state ->

  if (state.shotguns == 2 && state.score > 0) "stop"

}And my best idea benchmarks at an average of 2.6. Not much variation, really.

Jetty 7.4 Embedded Examples
2011-08-15T15:15:00.000-07:00
If you're finding that the Embedded Jetty documentation is out of date, then take a look at the source code for the Jetty examples in git:

$ git clone git://git.eclipse.org/gitroot/jetty/org.eclipse.jetty.project.git

$ cd org.eclipse.jetty.project/example-jetty-embedded/

Mac OSX Time Machine Backup to a NAS
2011-07-27T21:13:00.000-07:00
This guide will explain all the steps necessary. I stuck with a SMB share rather than hack my NAS (DNS-343) to accept the AFP (Apple File Protocol). If you're like me, you didn't have enough space on your hard drive left to create the initial time machine image. (Why they can't create it directly on the mounted network folder is a mystery to me). In case the guide disappears, here's the essence of it:

$ ifconfig en0 | grep ether
$ hdiutil create -size 140g -fs HFS+J -volname "Backup of MACHINENAME" MACHINENAME_MACID.sparsebundle

E.g.:
$ hdiutil create -size 140g -fs HFS+J -volname "Backup of magpie" magpie_0016cbaf91d7.sparsebundle

Then:

$ cp magpie_0016cbaf91d7.sparsebundle PATH_TO_NETWORK_MOUNT
$ defaults write com.apple.systempreferences TMShowUnsupportedNetworkVolumes 1

All of this will mean that Time Machine will skip the preparing step once you select your network mount and start the backup.

API Hackday
2011-07-27T11:00:00.000-07:00
I'll be at API Hackday PDX. Are you? If you didn't know already, this is what it's about:

"API Hackday PDX brings developers together the Saturday after OSCON for an all-day coding fest focused on building apps and mashups with APIs. Developers of all experience levels can share ideas, collaborate on existing projects, start new ventures, and find out about great tools and new APIs to play with."

12 hours of coding alongside other enthusiasts! What's not to like?

Compiling Gwt Modules Using Gradle
2011-07-27T10:48:00.001-07:00
Compiling Gwt modules using Gradle is pretty easy. Here's a nice way to avoid duplication if you have more than one:
tasks.addRule('Pattern: <module>Gwt') { taskName ->
  if (taskName.endsWith('Gwt')) {
    task "$taskName" (dependsOn: 'classes', type: JavaExec) {
      inputs.dir sourceSets.main.classesDir
      outputs.dir "$buildDir/gen"
      main = 'com.google.gwt.dev.Compiler'
      classpath {
        [ sourceSets.main.java.srcDirs,
          sourceSets.main.runtimeClasspath,
          configurations.gwt ]
      }
      args 'com.myplace.mypackage.' + (taskName - 'Gwt'),
         '-war', "$buildDir/gen",
         '-localWorkers', '2',
         '-style', 'PRETTY',
         '-draftCompile'
    }
  }
}Note the configurations.gwt which is a clean way of adding dependencies that are only required by the GWT compiler. E.g.:
configurations {
  gwt
}

dependencies {
  gwt ( group:'javax.validation', name:'validation-api',
        version:'1.0.0.GA', classifier:'sources' )
}

See You At ScaleOutCamp?
2011-07-26T07:06:00.000-07:00
I'll be at ScaleOutCamp this evening. It's not too late to sign up for the free event. See you there!

Get Your Groovy On #2 - Graffiti
2011-07-26T07:02:00.000-07:00
I've just posted my second Get Your Groovy On screencast to groovycasts.org. It's an example of simple RESTful web services in Groovy using Graffiti. Take a look.

Getting Away With Bad Git Amend Behaviour
2011-07-23T11:23:00.001-07:00
Ahem. Ever do this?

$ git commit ...
$ git push
$ git commit --amend ...
$ git push
To git@github.com:curious-attempt-bunny/gradle-github-deployer.git
 ! [rejected]        master -> master (non-fast-forward)
error: failed to push some refs to 'git@github.com:curious-attempt-bunny/gradle-github-deployer.git'
To prevent you from losing history, non-fast-forward updates were rejected
Merge the remote changes (e.g. 'git pull') before pushing again.  See the 'Note about fast-forwards' section of 'git push --help' for details.

BOOM! Whoops, yes, you forgot that you shouldn't be amending commits that you've already pushed to GitHub!

Here's the hackster way out (very very bad if anyone happens to have pulled or pushed in the interim):

$ git push origin +master:master

Gradle - Could not find implementation class 'null' for plugin
2011-07-23T10:40:00.000-07:00
I'm having fun hacking together a GitHub deployer plugin for Gradle. I just saw this error message. This means that there's a typo in your class name in your META-INF/gradle-plugins/PLUGIN_NAME.properties:

Caused by: org.gradle.api.plugins.PluginInstantiationException: Could not find implementation class 'null' for plugin 'githubDeployer' specified in jar:file:/Users/user/Documents/fun/gradle/gradle-github-deployer/plugin/build/libs/gradle-github-deployer-hack-0.1-SNAPSHOT.jar!/META-INF/gradle-plugins/githubDeployer.properties

Chess Board Computer Vision
2011-07-22T16:52:00.000-07:00
This is a topic that I keep coming back to. This time for Chess. The goal is to extract chess notation for a game played in view of a webcam. This time I made it harder for myself by viewing the board from an angle. I wrote it in Groovy using platform-independent JAI (Java Advanced Imaging). I've cheated on the webcam capture, and used Photo Booth stills to begin with. This time, though, I'm sharing the code on GitHub.

For each frame I do the following:
1. Convert to grayscale using "bandcombine":

2. Correct for perspective using PerspectiveTransform.getQuadToQuad and "warp": (I noted the pixel coordinates for the corners of the board by hand)

3. Binarize upper and lower bands of the image histogram and OR them together (so that a white pixel ideally identifies the presence of a piece on the board) -- using "histogram", "binarize", OrDescriptor, and SubtractDescriptor:

Then I take each processed frame and its following processed frame, and do the following:
1. XOR the two frames using XorDescriptor:

2. Erode the result using "erode":

3. Open it using "dilate" and then "erode":

The result would ideally show clusters of pixels in the board squares that have changed (e2 to e4 in this example). Unfortunately, a naive solution here isn't enough as pieces span multiple board locations. Also the piece detection algorithm in step 3 of frame processing is too weak.

Feel free to ask questions if you're interested.

Groovy Screencasts
2011-07-10T09:15:00.001-07:00
This is an idea I've been wanting to act on for a while. There aren't a lot of screencasts around on Groovy. With that in mind, I've created groovycasts.org and to kick things off I've recorded Get Your Groovy On. It's hard to cover much in 5 minutes, so this overview doesn't linger to go into detail. I'll gladly deep dive on a topic, though. What's of most interest to you? Let me know.

Here's the screencast (now with corrected audio):

SAARAA Wins First Place at RHOK 3
2011-06-05T23:35:00.000-07:00
Our SAARAA project for Seattle RHOK #3 won first place! Woot!

Participated in RHOK #3 Seattle
2011-06-05T16:01:00.000-07:00
I'm having a productive and engaging time here at RHOK #3 Seattle. I'm on the SAARAA (Situational Awareness And Rapid Assessment App) team. We've put together a client-server app using the Android SDK, Sinatra, and MongoDb.

Our github repositories are here:
* server-side
* client-side

Updating a Forked Git Repository
2011-04-30T10:36:00.000-07:00
I keep forgetting how to do this and have to hunt for the documentation. Here's the solution:$ git remote add upstream git://github.com/teamrabu/rabu-schedule.git
$ git fetch upstream

I'm on the SAO QA Forum Panel June 2nd
2011-04-29T19:40:00.000-07:00
I'm on the Panel for the Software Association of Oregon QA Forum June 2nd. There's a fee to attend:

Agile Death Match – Is QA obsolete in Agile?

Agile turns 10 years old this year and for 9.999 years we have struggled to understand QA’s role in a Agile environment.  Come join us as some of the region’s top Agilest and Quality professionals share their real world experience implementing QA in a agile world.

This lunch time presentation is ideal for the QA Engineer, Software Developer or IT leader who is trying implement or improve quality practices in a Agile team.

The intent is not to promote or redefine the Agile model, but rather to discuss the various ways the QA role has been impacted and has adjusted to the current model.
Please join us for this informative and interactive discussion followed by a Q&A session.

Test First Challenge - Modernized For The Lazy!
2011-04-03T20:04:00.000-07:00
Did you ever try the original Test First Challenge? You should have!

So what have I done with it?

* I've taken the tests and converted them to Spock specifications
* A Gradle build script creates your IDE setup
* The Gradle will add each new test as your code passes them

Check out the test-first-challenge project on github. The project is ready to go for either a Java or Groovy solution to the challenge.

Leave feedback here. Repository forks and pull requests welcome!

Free Groovy Training Workshop in Portland
2011-03-11T22:33:00.000-08:00
Hey there!

In preparation for Open Source Bridge 2011, I'm offering a Groovy training workshop at not cost to:motivated Java developers,
from the Portland community,
with little-to-no Groovy experience
in exchange for feedback for improving the session.

It'll be two hours long, collaborative, practical, and above all: fun! I haven't set a date yet. Follow up with me at: curious dot attempt dot bunny at gmail dot com with a subject line of Free Groovy Training Workshop.

Cheers,
Merlyn

Groovy For Java Developers - My Open Source Bridge Session
2011-03-09T21:31:00.000-08:00

I've submitted a session to Open Source Bridge – June 21–24, 2011 – Portland, OR. Please comment on my proposal for a Groovy For Java Developers session. Here's the teaser - follow the link for the full description:

Groovy is a powerful and concise language for the JVM that extends the capabilities of Java. Idiomatic Groovy can be intimidating to Java developers. This session will break the ice and reveal some of the power of Groovy as well as some of the traps.

Developing & Testing Gradle Plugins
2011-02-23T09:05:00.000-08:00
Here's a quick lazy post from me. I've been working on a Gradle OneJar plugin, and struggled along the way. This is a post to the gradle-user mailing list that I think it particularly helpful:Thanks for this, Peter. It's very helpful information.

On Wed, 2011-02-23 at 08:35 -0800, Peter Niederwieser wrote: 
> > I have a plugin I'm working on, and as far as I can tell I can't make a
> > multiproject build where one subproject consumes the plugin artifact
> > generated by the other subproject.
> > 
> 
> I think that putting the plugin artifact on the other script's class path
> (with a buildscript {} section) should work.

Hmm. It sounds like you're saying that the evaluation of subproject A
can depend on the execution of a task in subproject B. I didn't think
that was possible. 

> > Perhaps the gradle developers could shed some light on this. Which
> > versions
> > of the gradle java and groovy plugins does the gradle project itself
> > consume
> > during a build?
> > 
> 
> Before Gradle's integration tests are run, a full distribution is built,
> which will then be used by the integration tests. So the tests will always
> use the latest Java and Groovy plugins.

So for me there's good news and bad news. The bad news is that there's
no magic to make this a simple matter of configuring my build script.
The good news is that there are some gradle-core integration test
support classes.

For anyone else wanting to integration test their plugins, take a look
at the org.gradle.integtests package in gradle-core - particularly
org.gradle.integtests.fixtures. And for examples of their usage look at
the gradle integ-test subproject.

Thanks again, Peter!
Merlyn

My Agile Open NW 2011 Brain Dump
2011-02-10T09:00:00.000-08:00
These are potential talking points from AONW that I'm happy to discuss further:
Sessions I hosted: (flipchart photos uploaded)Getting To Thin Releasable Slices & Working Tiny-er
How Meta-Programming Is Changing The Way You Think
AONW Feedback Via The Perfection Game
Improving Your Facilitation Skills
Sessions that made a big impression:Core Protocols "Software For Your Head"
Coaching Dojo
Imago Dialogue
Maxims I noted down:(on process obstacles) "No precious resources" ... (regarding the solution) "Make it repeatable"
(on pairing) "We all have strengths in different areas. We can lead in different areas. Let's talk about it"
(on stunting leadership growth) "Find excuses to be out of the room"
(on talking about a sensitive topic) "What's the worst thing that could happen?"
Ideas I noted down:Use CI Servers to build VM Images with complete deployments
Coaching: Vent vs Help / Talk vs Answer
Coaching:Paraphrase
Ask from more (how is it?)
What have you tried
Characterize the situation
How do others perceive it
Not being emotionally involved or taking sides
Pausing
Mirroring what's understood
Mirroring body language
Mirroring pace of speech
Edit: Jeremy Lightsmith pointed me at Group Pattern Language.

Agile Open Northwest 2011 Is Over
2011-02-09T19:48:00.000-08:00
It's over! I put all the material I captured on the wiki page.

Installing Windows7 On A Machine Without Windows On It
2011-02-06T20:12:00.000-08:00
Okay. You'd think that Microsoft would make it easy for someone to convert to Windows, wouldn't you? Well, in fact, they don't. A friend was struggling to install Windows7 after paying for it online and downloading the installation files. I asked "What files do you have?", expecting an answer like "an ISO", or "an IMG". But no.. "An EXE, and two BOX files"!

And things just got crazier after that. If this is the problem you're facing, then you'll want to start here.

Before you do, know this: You absolutely, positively, need a Windows machine to get this working. Mind-blowing. Simply amazing.

You will need to download the Windows Installation Kit. But you can't do that on any other kind of OS. You need to download it on a machine that can run a program to prove it's a Genuine Windows installation. Nice.

Once you get that you'll have an IMG you can burn to DVD.

Once you install the software on the DVD you'll need to trawl through the installation directory for Oscdimg.exe.

Now you run your original EXE to extract data from the BOX files.

From a Windows CMD prompt you can then type the incantation for Oscdimg to produce an ISO image for a DVD.

You can then burn this ISO to a DVD, and you're good to go (it's possible that for this you may need to pay for some software that can burn ISOs to DVD).

Seriously, that's how easy it is?

PDX Software Development Site Offline
2011-02-04T18:50:00.000-08:00
As of today http://pdx-software-development.appspot.com/ is getting this exception talking to the Google Data API:Caused by: com.google.gdata.util.ServiceForbiddenException: OK
Maps Data api is now deprecated.
 at com.google.gdata.client.http.HttpGDataRequest.handleErrorResponse(HttpGDataRequest.java:593)
 at com.google.gdata.client.http.GoogleGDataRequest.handleErrorResponse(GoogleGDataRequest.java:563)
 at com.google.gdata.client.http.HttpGDataRequest.checkResponse(HttpGDataRequest.java:550)
 at com.google.gdata.client.http.HttpGDataRequest.execute(HttpGDataRequest.java:530)
 at com.google.gdata.client.http.GoogleGDataRequest.execute(GoogleGDataRequest.java:535)
 at com.google.gdata.client.Service.getFeed(Service.java:1101)
 at com.google.gdata.client.Service.getFeed(Service.java:972)
 at com.google.gdata.client.GoogleService.getFeed(GoogleService.java:617)
 at com.curiousattemptbunny.server.CompanyServiceImpl.refreshCompanies(CompanyServiceImpl.java:82)
 ... 37 more
For the time being you can refer to the underlying Google Map.

Grouch: A DSL For CouchDb
2011-01-23T18:45:00.000-08:00
I put together a simple DSL for interacting with Groovy today. I'm calling it Grouch*. Using Grouch, using CouchDb is as easy as this:import com.curiousattemptbunny.grouch.*

grouch = new Grouch('http://localhost:5984')
if (grouch.mydb.exists()) grouch.mydb.delete()
grouch.mydb.create()
grouch.mydb << [_id: 'mydoc', aField: 'aValue' ]
assert grouch.mydb.mydoc.aField == 'aValue'* Thanks to Ben Acker for the name. Funnily enough someone else had the same idea for a name but never pushed any code.

Driving Your Coding
2011-01-23T13:45:00.001-08:00
I think it's easy to end up coding the wrong thing or to speculate too much. No disagreement so far, right?

Apply that thinking to TDD. What's driving your choice of test cases? If you're like me, then it's very easy to jump in somewhere in the middle of the design and go from there. That works. The problem here though is this is a form of speculation in it's own right, and we can easily be inefficient with our own time.

Steven Freeman and others advocate an acceptance test driven TDD cycle. My thinking aligns with theirs on this to a certain degree. Explaining this to someone that's new to the concept leads to the reaction "So I have a test that's always red [until I'm done working on the feature]?". Answering "Essentially, yes." doesn't help too much it seems.

So how about this. This is a fragment of my code for a potential open source API I'm toying with:$ cat src/test/groovy/com/curiousattemptbunny/grouch/MotivatingExample.groovy
package com.curiousattemptbunny.grouch

couchdb = new Grouch('http://localhost')
couchdb.mydb.delete()
couchdb.mydb.create()
couchdb.mydb << [_id: 'mydoc', aField: 'aValue' ]
assert couchdb.mydb.mydoc.aField == 'aValue
I like this for a number of reasons:
1. It's clearly distinct from a regular unit test
2. It's very concise
3. It will fail if I'm not done with my tests

I encourage you to try this yourself and let me know how it went.

Impressive Date Parsing for JavaScript
2011-01-19T11:29:00.001-08:00
I'm impressed with datejs technically and even more so with the way its' function is communicated. If you take a look at their main page you'll see exactly what I mean. I was oscillating between two date formats '1/2/11 4:21:09 AM' and 'Sun Jan 02 04:21:19 PST 2011' and they both just worked out of the box and I didn't have to download it to know I'd found what I needed. Very well done datejs folks!

Want To Inline JVM Bytecode Into Your Groovy Code?
2011-01-16T13:40:00.001-08:00
Do you want to inline JVM bytecode into your Groovy code? Groovy's ASTs allowed the groovy-bytecode-ast project to come into existence. For instance, this method could be instead defined using the code that follows:
public int fib(int i) {
   return i<2?i:fib(i-2)+fib(i-1);
}

@groovyx.ast.bytecode.Bytecode
int fib(int i) {
    iload 1
    iconst_2
    if_icmpge l1
    iload 1
    _goto l2
   l1
    aload 0
    iload 1
    iconst_2
    isub
    invokevirtual '.fib', '(I)I'
    aload 0
    iload 1
    iconst_1
    isub
    invokevirtual '.fib' ,'(I)I'
    iadd
   l2
    ireturn
}
println fib(40)

How is this done? Surprisingly, there's actually not very much in the way of code needed to support this. The key implementation details are in BytecodeASTTransformation.groovy - a Local AST Transformation. Way to go Cédric Champeau!

Update: now with new groovified syntax and matching IntelliJ ASM plugin.

I'm Mentoring At Agile PDX's Tune-Up Workshop Jan Wednesday 19th
2011-01-15T10:36:00.000-08:00
In case you're not following AgilePDX on Twitter or keeping up to date with the [now poorly named] xpportland mailing list:

This event on Calagator.

This month, come to the Agile PDX 'Agile Tune-Up' meeting and tune up your agile process! Our panel of experts and practitioners will give you personalized feedback on the issues you're facing in an informal workshop-like format.

Before the session, visit http://agile-pdx-tune-up.appspot.com/ for a brief questionnaire that will help you identify your areas of greatest risk. Then bring a printout with you to the session to get targeted advice on what to keep doing and what to change!

This event is free and is at our new location, Puppet Labs. It begins at 6:30 pm with pizza, sponsored by PNSQC (Many thanks to both Puppet Labs and PNSQC for supporting agile in Portland).

The program starts at 7:00 pm.

After the program you're invited to join us for a no-host gathering at a nearby brewpub for further discussion.

I'm Facilitating At PJUG January Tuesday 18th
2011-01-15T10:33:00.000-08:00
This session isn't Java specific as it applies equally well to Groovy and Scala, so if you're not on the PJUG list you may miss this:

Hi,

I'm Merlyn, a Portland-based programmer with strong ties to both Agile
PDX and Groovy. I'm also a member of an Agile team at YesMail, and I
blog under the handle "curious attempt bunny".

Come to the session if you want to participate in learning more about
Gradle. Gradle is a powerful layer of Groovy built on top of a number
of established tools and libraries. Smart guys have borrowed from the
pioneering work of Ant, Maven, Gant, Buildr and come up with an
expressive, concise, and convention-rich build tool for the Java
family.

For this session, there won't be any slides. I'm also not going to
stand up and talk at you. Instead, I'm going to rely heavily on your
involvement. I'll start with an introduction to Gradle, and then very
quickly go into using it in practice. After a few minutes of that I'll
open it up to go wherever we're most interested in. I'm not a Gradle
expert, yet I believe that my Groovy background in conjunction with a
few days of earnest usage have taught me a lot. This session will rely
heavily on an internet connection, and we'll be googling for answers
together if need be.

Do this ahead of time:

What is your intention for the session? Spend some time deliberately
focusing on what it is you want out of the session. Bring that with
you and share it with us. It will help us all focus on what's
important and go some way to helping you get the most out of the
session. Also, go have a quick read about Gradle on their website.
That will save us all some time, help us get to the interesting
details sooner, and be a good source of informed questions for you.

I'll also be building in some time at the end of the session to gather
feedback that will help me learn to become a better facilitator.

Cheers,
Merlyn

PS Regardless of whether you come to the session, if you practice
Agile at work, or are struggling to do so. then do plan to come to the
Wednesday 19th Agile PDX Tune-Up Workshop where I and a number of
local practitioners will be mentoring alongside James Shore. More
details here:  http://calagator.org/events/1250459616

Print Previw in GWT
2011-01-11T06:51:00.000-08:00
This took me a while to find so I thought I'd share it with you. I'm assuming you're using a solution like this in your HTML:

<link rel="stylesheet" type="text/css" media="print" href="Print.css">

But what if you want the user to view the print preview inside the browser? In that case you can dynamically include your css like so:

StyleInjector.inject("@import url(\"Print.css\");

This doesn't address how to make this toggleable - that's for you to find out (and let me know how you did it). In my case I can simply reload the app:

Window.Location.reload();

It Just Makes Sense
2011-01-09T19:06:00.000-08:00
subscribe.groovyrss = new URL("http://day9tv.blip.tv/rss").text
root = new groovy.util.XmlSlurper().parseText(rss)
root.channel.item.each { item ->
  def url = item.enclosure.@url.text()
  def name = new File(url).name

  if (!new File(".$name").exists()) {
    println "Downloading $name : $url"
    new File(".$name").write("")

    def file = new File(name).newOutputStream()  
    file << new URL(url).openStream()  
    file.close() 
  }
}

Night and Day - Working at the Right Place
2011-01-09T08:27:00.000-08:00
This is an extract from a user group post someone made. What I'd like to highlight about it is specifically that you don't really know how bad you had it until you find somewhere heathy. And to managers and employers: a sane/humane working environment where you treat people with respect is invaluable (a medium-term requirement?).

Hi, [appologies for my earlier rant which must be] a symptom of my latent anger, at the lousy way [CompanyX] treated myself and some of my friends during the last decade.

It  was brought  into perspective  because I am  so much happier in my current position, working with a  small group of software geeks, who all enjoy the creativity and attention to detail  that it takes  to write (and sell) our software.  So  my salary  is basically the same as it was  10 years ago ,  and the benefits aren't quite as comprehensive, but being empowered and able to speak out (constructively) and working with a  smart manager and smart bunch of people, * more than* makes up  for that.
*SOME ADVICE:*
1. You should be very careful  about what you sign up for, and *who you work with  and how they treat you*.  Don't just think about the  $$  and benefits. Look around the office  and talk to some of the developers (without anyone looking over your shoulder).
2.* Look after your body and brain*. I  didn't know this until a few years ago,  but if you have to work where there is limited natural light , and you feel ill a lot,  see your  Doc  to get some  SSRI  'happy pills'. Many people get depressed or anxious because the levels of Serotonin in their Brains drops when they *have to work in the gloom*. Our profession is  *very sedentary* ...you need to keep moving  and walk or exercise ...otherwise you will get  out of shape  and that can lead  to some *very  *nasty problems. *Do not* sit at your desk all day!
3. Germs and viruses spread like wildfire when you are  *cooped up in an office*. If your neighbor insists on  'sharing' their flu with you ... tell them to go home and take one of their sick days (if you get sick days) ..and remember to wash your hands before you eat,  or keep a tub  of Lysol wipes handy.
4. IF you're not happy with the company or your manager, *don't take the crap*,  like I did,  find another job. It will be scary but you'll be so much wiser and happier!

So, how good do you have it?

Ubuntu 10.10 & HP Pavillion Erratic Touchpad
2010-12-16T10:42:00.001-08:00
Having trouble selecting from drop down boxes or selecting blocks of text? Does the mouse cursor go crazy if you touch the pad with two fingers? Does right click not work? Does your mouse cursor jump to the bottom of the screen when you try to left click? If so, this script from SansMicrosoft will fix your issues.

Shrinking Your Windows 7 Partition So That You Can Install Ubuntu 10.10
2010-12-15T15:44:00.001-08:00
Having trouble shrinking your NTFS partition on Windows 7 so that you can dual boot into Ubuntu 10.10? Windows7 Disk Manager sucked for me, suggesting that I could reclaim just 50% of my practically empty NTFS partition. What worked for me (and worked in literally 3 minutes) was Partition Wizard Home Edition. What's more, it's free to use!

Setting Up Gitosis on Mac OSX
2010-11-15T08:00:00.001-08:00
These appear to be the defacto instructions for setting up gitosis. And they worked very well for me. The only trouble I had was with the clone url. This form didn't work for me: git clone git@YOUR_SERVER_HOSTNAME:gitosis-admin.gitInstead I needed to use this:git clone git@YOUR_SERVER_HOSTNAME:~/repositories/gitosis-admin.gitEnjoy!

Grails, CometD, and Lazy Initialization of Services
2010-11-14T10:48:00.000-08:00
I've been giving CometD a go in Grails. There's a good how-to on that. The problem I encountered immediately is that Grails doesn't eagerly construct my Service bean, so bayeuxSession.handshake() never gets called. After trying in vain to find a clean way to have Service beans eagerly initialized, I settled on this change to BootStrap.groovy:import org.codehaus.groovy.grails.commons.ApplicationAttributes

class BootStrap {
  def init = { servletContext ->
    def ctx =
      servletContext.getAttribute(ApplicationAttributes.APPLICATION_CONTEXT) 
    ctx.getBean("lobbyService")
  }

  def destroy = {
  }
}Incidentally, if you're not finding the CometD 2.0.0 javadocs, it's because they're not online. Instead, grab the docs from maven - e.g. http://repo2.maven.org/maven2/org/cometd/java/bayeux-api/2.1.0.beta2/bayeux-api-2.1.0.beta2-javadoc.jar.

Writing a Game for the Android
2010-11-05T15:50:00.000-07:00
I recently bought myself a G2 Android phone, and being a developer the first thing I had to do was write an app for it. Here are some of the useful tips I learnt along the way:
Use a SurfaceView for graphics-biased apps
See the LunarLander SDK example. You'll want to getHolder(), listen for SurfaceHolder.Callback events, and create a worker thread to control the rendering.
Listen to touch events
The SurfaceView can register for onTouch events using setOnTouchListener(this).
Draw bitmaps
Load resources using BitmapFactory.decodeResource(res, R.drawable.someBitmap) and then canvas.drawBitmap will do the job. For transparencies create a Paint and call paint.setAlpha(alpha).
Scale, translate, and rotate the canvas
canvas.save() and canvas.restore() are your friends along with methods such as canvas.scale(scaleX, scaleY).
Make the app fullscreen
Put this in your manifest.xml for the Activity: android:theme="@android:style/Theme.NoTitleBar.Fullscreen"
Lock the orientation of your app
Put this in your manifest.xml for the Activity: android:screenOrientation="portrait"
Pop up dialogs - but not from the render thread!
AlertDialog.Builder ad = new AlertDialog.Builder(context);
ad.setTitle("You won!");
ad.setMessage("Play again, sir!");
ad.setPositiveButton("New Game", new OnClickListener() {
  public void onClick(DialogInterface dialog, int which) {
    newGame();
  }
});Play sounds
MediaPlayer mp = MediaPlayer.create(context, R.raw.someSound))); and then later: mp.start();
Make the volume controls change the media sounds and not the ringer
setVolumeControlStream(AudioManager.STREAM_MUSIC);

Enjoy!

Refreshingly Concise Builds
2010-10-24T11:14:00.000-07:00
Gradle has a direct tag line: "A better way to build". But is it really?

To start with, Gradle's build configuration files themselves are very concise, e.g.:
apply plugin: 'java'
apply plugin: 'eclipse'

sourceCompatibility = 1.5
version = 0.1

repositories {
  mavenCentral()
}

dependencies {
  compile (
    "asm:asm-all:3.3",
    "org.slf4j:slf4j-api:1.6.1"
  )
  testCompile (
    "junit:junit:4.8.2"
  )
}Gradle isn't ashamed to take back information it's shown you. While it's running a build you'll see output relevant to it's current task, but if the task runs without issue it'll hide all the noise and give you just the overview.

While in progress:
> Building > :compileJavaWhen all goes well:
$ gradle test
:compileJava UP-TO-DATE
:processResources UP-TO-DATE
:classes UP-TO-DATE
:compileTestJava UP-TO-DATE
:processTestResources
:testClasses
:test

BUILD SUCCESSFUL

Total time: 8.149 secsContrary to other build tools, when things do not go well the output is verbose enough to be helpful while avoiding a scatter-shot let's-dump-out-everything-that-was-going-on approach.

Java 1.7 ... Sometime?
2010-09-08T20:29:00.000-07:00
First read this and capture your reaction: Java 1.7 Release Date.

Was it anything like this?

"Wow. Just... wow.

2012. Really? /Really/?"

:D

Java 1.7: What's In The Pipeline
2010-08-31T08:40:00.000-07:00
Here's what will probably be going into Java 1.7 from Project Coin.

For those who can't upgrade from Java 1.5, you may be able to find them in Groovy:

"Supporting some of these JDK 7 aspects on JDK 5+ is also a nice appeal
to Java developers who can't just upgrade"

Help Me Solve My Google Maps API Info Window Problem
2010-07-24T11:09:00.000-07:00
I'm quite stuck. I'm using the Google Maps API via the gwt-google-apis project. The symptom of the problem is that I create a <table> (that Firebug says has computed dimensions of 658 x 196) and Google Maps API wraps this with a <div style="position: absolute; left: 16px; top: 16px; width: 658px; height: 142px; z-index: 10;">.

I haven't found any way to force the info window to be resized or to set it's height. This is what it looks like:

Notice how the edit button floats off the bottom of the bubble. Yuck!

Here's the link to the app (click the YesMail link after the list has loaded).

Any ideas? I'm using this method to display the info window:
FlexTable content = new FlexTable();
//...
map.getInfoWindow().open(company.getMarker(),
  new InfoWindowContent(content));

Groovy the #1 JVM scripting language
2010-07-22T22:02:00.000-07:00
"JVM scripting language No. 1: Groovy -- Powerful, high-level, enterprise-friendly"

com.google.gdata.util.ParseException: name must have a value
2010-07-20T21:44:00.000-07:00
For some reason the Google Maps API gdata core recently started failing to read my feature feed. This is the error I see in my App Engine log:
Caused by: com.google.gdata.util.ParseException: [Line 1, Column 3457, element atom:name] name must have a value
 at com.google.gdata.util.XmlParser.throwParseException(XmlParser.java:730)
 at com.google.gdata.util.XmlParser.parse(XmlParser.java:693)
 at com.google.gdata.util.XmlParser.parse(XmlParser.java:576)
 at com.google.gdata.data.BaseFeed.parseAtom(BaseFeed.java:867)
 at com.google.gdata.wireformats.input.AtomDataParser.parse(AtomDataParser.java:68)
 at com.google.gdata.wireformats.input.AtomDataParser.parse(AtomDataParser.java:39)
 at com.google.gdata.wireformats.input.CharacterParser.parse(CharacterParser.java:100)
 at com.google.gdata.wireformats.input.XmlInputParser.parse(XmlInputParser.java:52)
 at com.google.gdata.wireformats.input.AtomDualParser.parse(AtomDualParser.java:66)
 at com.google.gdata.wireformats.input.AtomDualParser.parse(AtomDualParser.java:34)
 at com.google.gdata.client.Service.parseResponseData(Service.java:2165)
 at com.google.gdata.client.Service.parseResponseData(Service.java:2098)
 at com.google.gdata.client.Service.getFeed(Service.java:1136)
 at com.google.gdata.client.Service.getFeed(Service.java:998)
 at com.google.gdata.client.GoogleService.getFeed(GoogleService.java:631)
 at com.google.gdata.client.Service.getFeed(Service.java:1017)
 at com.curiousattemptbunny.server.CompanyServiceImpl.refreshCompanies(CompanyServiceImpl.java:80)
 ... 52 more
Caused by: com.google.gdata.util.ParseException: name must have a value
 at com.google.gdata.data.Person$AtomHandler$NameHandler.processEndElement(Person.java:251)
 at com.google.gdata.util.XmlParser.endElement(XmlParser.java:1004)
 at org.xml.sax.helpers.ParserAdapter.endElement(Unknown Source)
 at com.sun.org.apache.xerces.internal.parsers.AbstractSAXParser.endElement(Unknown Source)
 at com.sun.org.apache.xerces.internal.parsers.AbstractXMLDocumentParser.emptyElement(Unknown Source)
 at com.sun.org.apache.xerces.internal.impl.XMLNSDocumentScannerImpl.scanStartElement(Unknown Source)
 at com.sun.org.apache.xerces.internal.impl.XMLDocumentFragmentScannerImpl$FragmentContentDriver.next(Unknown Source)
 at com.sun.org.apache.xerces.internal.impl.XMLDocumentScannerImpl.next(Unknown Source)
 at com.sun.org.apache.xerces.internal.impl.XMLNSDocumentScannerImpl.next(Unknown Source)
 at com.sun.org.apache.xerces.internal.impl.XMLDocumentFragmentScannerImpl.scanDocument(Unknown Source)
 at com.sun.org.apache.xerces.internal.parsers.XML11Configuration.parse(Unknown Source)
 at com.sun.org.apache.xerces.internal.parsers.XML11Configuration.parse(Unknown Source)
 at com.sun.org.apache.xerces.internal.parsers.XMLParser.parse(Unknown Source)
 at com.sun.org.apache.xerces.internal.parsers.AbstractSAXParser.parse(Unknown Source)
 at com.sun.org.apache.xerces.internal.jaxp.SAXParserImpl$JAXPSAXParser.parse(Unknown Source)
 at org.xml.sax.helpers.ParserAdapter.parse(Unknown Source)
 at com.google.gdata.util.XmlParser.parse(XmlParser.java:685)
 ... 67 more
The culprit appears to be the lack of an <atom:name> tag in the feature feed. Until the reason for this reveals itself or an update to the API is released, I'm using a custom build of the gdata-core-1.0.jar. I simply substituted a non-null value for the name in the Person.NameHandler class:
if (value == null) {
  //throw new ParseException(CoreErrorDomain.ERR.nameValueRequired);
  value = "N/A";
}

Sun/Oracle - You Suck
2010-07-20T21:12:00.000-07:00
require.client.dependencies:

BUILD FAILED
/.../gdata-java-client/java/build-src/core.xml:394: missing jar file: /tmp/gdata_dep/mail.jar
The GData client requires Sun's javamail API (version 1.4), which
is not included in this distribution.

You can download it from:
  http://java.sun.com/products/javamail/downloads/index.html

Then save it under:
  /tmp/gdata_dep/mail.jar

Time Machine Backup Over The Network
2010-07-09T19:40:00.000-07:00
Are you like me and want to back up just a few pet projects to another one of your computers on the network? If so it's not hard:

On the machine you want to backup to:
* Share a new folder in the System Preferences > Share page.

On the machine you want to backup from:
* Open up the shared folder in Finder: Go > Network.
* From Terminal run:
defaults write com.apple.systempreferences \
TMShowUnsupportedNetworkVolumes 1
* Open up Time Machine Preferences and exclude everything - you can shift-select everything on your HDD. Then selectivels remove from the list the things you don't want excluded - if you follow me.
* Verify that the total calculated size is managable - less than 10Gb say.
* Kick off a backup!
* Come back in a day or so! :)

Edit: My 9.7Gb backup over 11g wireless took just under 3 hours.

MapsService setUserCredentials Hang Problem
2010-07-08T07:05:00.000-07:00
Is the Google Data Maps Feed API MapsService.setUserCredentials hanging for you? You're probably using Java 1.5. Switch to Java 1.6 and the problem goes away. 

Best summary of the problem:

"I'm seeing the same thing at the same spot. CharMatcher enters into an infinite loop in setBits() while initializing the SINGE_WIDTH static instance. However, it does NOT hang if I single step over, or through, the initialization in the Eclipse debugger. I can reproduce this hang outside of the setUserCredentials() method by simply forcing the CharMatcher class to be loaded."

Solution found here:

"A follow-up. I was able to switch back to using a 1.6 JVM (1.6.0_18) and found that the hang no longer occurs. Something must be screwed up with how a 1.5 JVM handles the byte code that was generated. Perhaps recompiling the code with a different version of the javac would change the behavior?"

See gdata-client issue 249.

Grease Monkey & Cooks Illustrated Recipe Printing
2010-07-07T21:31:00.000-07:00
Install this script I wrote. Then:

Navigate to a recipe on CooksIllustrated.com. Click the Print link. Now click the Shrink button to eliminate the noise and have a recipe that will print to the width of an index card. 

To illustrate, here are the before and after screenshots:

Before

After

This is the code (also here):

// ==UserScript==
// @name           CooksIllustrated.com - shrink recipies to index card width for printing
// @namespace      http://curious-attempt-bunny.blogspot.com/
// @include        http://www.cooksillustrated.com/recipes/print/*
// @require        https://ajax.googleapis.com/ajax/libs/jquery/1.3.2/jquery.min.js
// ==/UserScript==

$('body').append('<div id="gm_shrink">shrink</div>');

$('#gm_shrink').css(
 'position', 'fixed').css(
 'top', '2px').css(
 'right', '2px').css(
 'background-color', '#991111').css(
 'color', 'white').css(
 'cursor', 'pointer').css(
 'padding', '2px 4px 2px 4px').css(
 'border', 'solid 1px #000000').css(
 'font-weight', 'bold').click(function() {
 var shrink = function(el) {
  el.css('font-size', '8px').css('margin', '0')
   .css('line-height', 'normal').css('font-size', '8px');
 }

 $('body').css('width', '310px');
 $('.pageSectionTop').after('<p><b>'+$('h1').text()+'</b></p>');
 $('b').css('font-size', '12px');
 $('.dek').hide();
 shrink($('em'));
 shrink($('p'));
 shrink($('td'));
 shrink($('a'));
 $('.pageSectionTop').hide();
 $('.pageSection').hide();
 $('img').hide();
 $('br').hide();
 $('td').css('padding', '0');
 $('#contentFooter').hide();
 $('.footerText').hide();
 $('#gm_shrink').hide();
});

That Doesn't Happen Very Often!
2010-07-07T19:45:00.001-07:00

Fifty Dangerous Things
2010-07-07T19:27:00.001-07:00
Continuing on from Learning as Fun: 50 dangerous things (you should let your kid do)

What a perfect book! Some of my favourites:
* Play with the vacuum cleaner
* Drop from high places
* Break glass
* Deconstruct an appliance
* Poison your friends
* Super glue your fingers together
* Make a bomb in a bag
* Boil water in a paper cup

AdSense and Gwt Apps?!?
2010-07-06T21:12:00.000-07:00
So I have a GWT app with "zero content" because all the content is rendered by GWT. How do other people solve this?

From Google AdSense:

"Thank you for your interest in Google AdSense. Unfortunately, after
reviewing your application, we're unable to accept you into Google AdSense
at this time.

We did not approve your application for the reasons listed below.

Issues:

- Under construction

---------------------

Further detail:

Under construction: We've found that your site's pages are incomplete or
under construction. To be considered for AdSense, we require websites to
be fully launched and functioning, allowing users to navigate throughout
your site with a menu, sitemap, or appropriate links. Your site must also
contain enough content for our specialists to review and to display
relevant ads. Once the majority of your site is complete and functional,
we'll be happy to reconsider your application. 

Learning as Fun
2010-07-06T21:10:00.000-07:00
Just add points? What UX can (and cannot) learn from gamesView more presentations from Sebastian Deterding.

Updated: Places To Work Down-Town In Portland, OR
2010-07-05T10:37:00.000-07:00
I've updated the map of the places a software developer might like to consider working at within a 2 mile radius of down-town Portland. I've changed the Collaboration settings to allow anyone to edit the map:

View Larger Map / See Full List

How Agile Are We?
2010-07-04T00:26:00.001-07:00
Assess your team's Agility, try out this app I've written - and give me some usability feedback!

The app is hosted on appengine: http://how-agile-are-we.appspot.com/.

All the material comes from The Art of Agile Development.

Assess Your Agility - The Lazy Way
2010-06-27T23:26:00.000-07:00
I've put together a little GWT application based around the Assess Your Agility section of The Art of Agile Development. It's work in progress using gwt-graphics:

The idea is that it asks you to answer the questions first that will make the biggest difference to your agility assessment. That lets you stop as soon as you get tired of it, whilst giving you a full a picture as possible. Ideally you'd accumulate a weighted "to-improve" list of Agile Practices.

A Cure For The Common Cold?
2010-05-31T22:05:00.000-07:00
It's not a cure, but it most definitely is a combination of rarely* found together ingredients:

#1 Celestial Tea's Wild Berry Zinger
#2 Ribena
#3 Jack Daniels

This tastes good, but far more likely to be of use is a Neti Pot.

* #1 and #2 being indigenous to separate continents.

Configuring TextWrangler To Edit Git Commit Comments
2010-05-29T14:45:00.001-07:00
git config core.editor "edit -w"

Heuristics For Solo Dominion Play II
2010-05-27T22:15:00.000-07:00
Next steps are to calculate values of different decks (to aid in choosing what to buy to add to them). Right now this is a money-only look at this calculation.

Not sure what I'm talking about? Read the previous post in this thread. 

As a quick test, what's the expected value of a starting hand?

Deck [Copper:7, Estate:3] has mean hand money value of 3.50 with variance 0.583
Distribution of hand values:
2: 0.083%
3: 0.42%
4: 0.42%
5: 0.083%

To test out some ideas. If it were possible to choose between buying 2 Silver or 1 Gold, which would be better?

Early in the game 2 Silver beat 1 Gold:

Deck [Copper:7, Estate:3, Gold:1] has mean hand money value of 4.55 with variance 1.88
Distribution of hand values:
2: 0.045%
3: 0.23%
4: 0.24%
5: 0.18%
6: 0.23%
7: 0.076%
Deck [Copper:7, Estate:3, Silver:2] has mean hand money value of 4.58 with variance 1.30
Distribution of hand values:
2: 0.027%
3: 0.15%
4: 0.29%
5: 0.32%
6: 0.17%
7: 0.044%

Later in the game 1 Gold beats* 2 Silver:

Deck [Copper:7, Estate:3, Silver:7, Gold:5] has mean hand money value of 8.18 with variance 3.88
Distribution of hand values:
2: 0.00080%
3: 0.0058%
4: 0.023%
5: 0.059%
6: 0.11%
7: 0.17%
8: 0.19%
9: 0.18%
10: 0.14%
11: 0.078%
12: 0.032%
13: 0.0093%
14: 0.0013%
15: 3.8e-05%
Deck [Copper:7, Estate:3, Silver:5, Gold:6] has mean hand money value of 8.33 with variance 4.32
Distribution of hand values:
2: 0.0010%
3: 0.0069%
4: 0.023%
5: 0.057%
6: 0.10%
7: 0.16%
8: 0.18%
9: 0.18%
10: 0.14%
11: 0.088%
12: 0.044%
13: 0.015%
14: 0.0037%
15: 0.00029%

* beats means here: it maximizes your expected hand value.

As a side note. The algorithm for calculating this data is a pretty interesting one. Give it a shot.

Test Data Builders Part II
2010-05-27T20:22:00.000-07:00
Following on from Test Data Builders Part I, one of the authors of Growing Object Oriented Software has produced a small DSL framework, called MakeItEasy, for writing test data builders. Once you have written the code for the builders, using them reads like this:

Maker anOldPerson = a(Person, with(age, 80));
doSomethingWith(make(anOldPerson));

Maker anOldWoman = anOldPerson.but(with(name, "Alice"));
Maker anOldMan = anOldPerson.but(with(name, "Bill"));
doSomethingElseWith(make(anOldWoman));
doWhateverWith(make(anOldMan));

This DSL makes for easy reading. It's still going to be a pain to write all the boilerplate code for the builders, though. Perhaps some intelligent application of Groovy Local ASTs, presenting Java-friend interfaces, could save the day?

Heuristics For Solo Dominion Play
2010-05-27T07:38:00.000-07:00
Have you tried Dominion yet? It's a lot of fun! Like all good strategy games Dominion just send me into a spin of "how would I code a computer to play this game?"

Here's how I've scaled down the game for very simple solo play:

1. Only money cards and the Province are purchasable
2. There are five Province cards

A quick overview of the rules: you start with a deck of seven Copper and three Estates (Estates are worthless in this solo play game as they don't help you win). You keep drawing five cards from your deck, shuffling you discards to make a new deck as needed. You sum up the purchasing power of each money card in your hand and optionally buy a card (0 for Copper, 3 for Silver, 6 for Gold, 8 for Province). The bought card and the hand of five cards then go into the discard. The game ends when the fifth and final Province card is purchased. The aim of the game is to finish after the fewest number of hands drawn.

So what's the best strategy for play in the very very simplified version of the game? You'd hope that it would be "buy the highest value card that I can with the money that I have".

It turns out that that's close, but there is better. I wrote a simulation for the game and coded up a number of different heuristics. In the style of Contract Bridge (essentially n games with fixed random seeds for each) I compared all the heuristics using a little ad-hoc parameter value tuning along the way. After that I ran a million-game simulation to compare them all. It turns out that the most effective strategy was to never ever buy copper coins (go without instead), to otherwise always buy the most expensive card you can with the exception of Province cards which you only buy on or after your eighth hand.

Howsat!?

I tried to come up with something more to do with the money-value distribution instead of the hand number but they performed worse. Here's the breakdown:

Calculation for a sample size of 1000000 games:
buyHighestPossible: 19.229789 average number of hands to finish
buyHighestPossibleBeforeRoundEight: 19.189278 average number of hands to finish
buyHighestPossibleBeforeRoundEightButNeverCopper: 19.086139 average number of hands to finish
buyHighestPossibleBeforeRoundEightButNeverCopperExceptInTheFirstTwoHands: 19.129784 average number of hands to finish
buyMoneyUntilAverageHandValueExceedsSixPointSix: 19.194762 average number of hands to finish

It's still a very small difference. You'd have to play ten games before you'd have a reasonable chance of noticing the difference between the #1 and #2 strategy.

Edit: Wow. With the averages being so similar the variance would be very useful information.

Calculating StarCraft2 Build Order Templates: Part II
2010-05-16T12:25:00.000-07:00
Wow. It works. Now that I've integrated a proper model of the economy this is pretty accurate! Here's what my program output:
Goals are: [prism, warp, blink]
Required plan elements are: [prism, warp, blink, robotics, cyber, gate, twilight, assim]

Searching...
Best so far: [gate, cyber, assim, warp, robotics, prism, twilight, blink]: 8:12
Best so far: [gate, assim, cyber, warp, robotics, prism, twilight, blink]: 7:25
Best so far: [assim, gate, cyber, warp, robotics, twilight, prism, blink]: 7:03
Best so far: [assim, gate, cyber, warp, robotics, twilight, blink, prism]: 7:01
Best so far: [assim, gate, cyber, warp, twilight, blink, robotics, prism]: 6:53
Best so far: [assim, gate, cyber, warp, twilight, robotics, blink, prism]: 6:48
There are 3 equivalently optimal plan(s):
[assim, gate, cyber, warp, twilight, robotics, blink, prism]: 6:48
[assim, gate, cyber, twilight, warp, robotics, blink, prism]: 6:48
[assim, gate, cyber, twilight, robotics, warp, blink, prism]: 6:48

Detail for [assim, gate, cyber, warp, twilight, robotics, blink, prism]:
@ 75 minerals (~8/10) -> assim (~0:32..1:02)
@ 150 minerals (~12/18) -> gate (~2:13..3:18)
@ gate (~16/18) -> cyber (~3:18..4:08)
@ cyber (~19/26) -> warp (~4:08..6:28)
@ 150 minerals, 100 gas (~19/26) -> twilight (~4:08..4:58)
@ 200 minerals, 100 gas (~19/26) -> robotics (~4:21..5:26)
@ twilight (~21/26) -> blink (~4:58..6:48)
@ robotics (~23/26) -> prism (~5:26..6:16)

Trying to follow this, my actual timings were:
8/10 (0:37) assim
8/10 (1:00) pylon
12/16 (2:12) gate
14/16 ?? pylon
16/16 (3:19) cyber
18/26 ?? warp
18/26 (4:20) twilight -- 10 seconds late
18/26 (4:25) robotics
24/26 ?? pylon
?? ?? blink
25/26 (5:33) prism -- 10 seconds late

I call that a success! Seriously, what were the chances that the simulation would be that accurate!?

Optimal Economy Strategies in StarCraft2
2010-05-15T21:59:00.000-07:00
The main problem I had with Calculating StarCraft2 Build Order Templates was having such an inaccurate model of the economy. I've made some progress on that front, and have posted my findings on the TeamLiquid strategy forum here.

Calculating StarCraft2 Build Order Templates
2010-05-11T22:47:00.001-07:00
So, you're interested in StarCraft2, and you have a plan for your game, e.g.: I want to attack using stalkers, with the blink upgrade, and ferry them to the enemy in a warp prism. Okay, but it could take a while to work out a sensible basic order to build these things in so that you can attack at the earliest possible time. This is where I apply my programming hammer to my StarCraft2-experience-deficit nail:// name, mineral cost, gas cost, time to build
cyber = new Item("cyber", 150, 0, 50)
warp = new Item("warp", 50, 50, 140)
robotics = new Item("robotics", 200, 100, 65)
prism = new Item("prism", 200, 0, 50)
twilight = new Item('twilight', 150, 100, 50)
blink = new Item('blink', 150, 150, 110)

warp.requires = cyber
robotics.requires = cyber
prism.requires = robotics
twilight.requires = cyber
blink.requires = twilight
I've started coding up something naive with a friend and have it outputting this (with lots of assumptions and simplifications, e.g. time starts at 0:00 with a gateway in existence, no new probes are made, arbitrary mining rate used, ignored gas cost):

0:38..1:28: cyber
2:06..2:56: twilight
2:19..4:39: warp
3:09..4:14: robotics
3:47..5:37: blink
5:04..5:54: prism

Critical path:
0:38..1:28: cyber
3:09..4:14: robotics
5:04..5:54: prism

If this pans out, it should lead to:

* an ideal order in which to build,
* build triggers / approximate food timing,
* the earliest a plan could be realized,
* the key components that being slow on would delay the plan.

Edit: We've made some more progress. This is more complete but is still missing a good model of the economy's probe building and behaviour:
Goals are: [prism, warp, blink]
Required plan elements are: [prism, warp, blink, robotics, cyber, gate, twilight, assim]

Searching...
Best so far: [gate, cyber, assim, warp, robotics, prism, twilight, blink]: 6:33
Best so far: [gate, assim, cyber, warp, robotics, prism, twilight, blink]: 5:58
Best so far: [gate, assim, cyber, warp, robotics, twilight, prism, blink]: 5:42
Best so far: [assim, gate, cyber, warp, robotics, twilight, prism, blink]: 5:22
Best so far: [assim, gate, cyber, warp, robotics, twilight, blink, prism]: 5:15
There are 6 equivalently optimal plan(s):
[assim, gate, cyber, warp, robotics, twilight, blink, prism]: 5:15
[assim, gate, cyber, warp, twilight, robotics, blink, prism]: 5:15
[assim, gate, cyber, robotics, warp, twilight, blink, prism]: 5:15
[assim, gate, cyber, robotics, twilight, warp, blink, prism]: 5:15
[assim, gate, cyber, twilight, warp, robotics, blink, prism]: 5:15
[assim, gate, cyber, twilight, robotics, warp, blink, prism]: 5:15

First plan detail:
@ 75 minerals -> assim (~0:05..0:35)
@ 150 minerals -> gate (~0:32..1:37)
@ gate -> cyber (~1:37..2:27)
@ cyber -> warp (~2:27..4:47)
@ 200 minerals, 100 gas -> robotics (~2:27..3:32)
@ 150 minerals, 100 gas -> twilight (~2:27..3:17)
@ 150 minerals, 150 gas -> blink (~3:25..5:15)
@ robotics -> prism (~3:32..4:22)

Experimenting With Haskell
2010-05-07T21:26:00.000-07:00
I tried implementing a simple solution in Haskell as many ways as I could think of. The problem:

Take a list of numbers and do something similar to run length encoding. So [1,1,2,2,2,4,3,3] should become [(1,2),(2,3),(4,1),(3,2)].

At the end are the solutions I came up with. I like the final solution the best. Are there other ways that you know of?

It bugs me that I can't define implicit guards so that instead of f a b | a == b = ... I could say f a a = ...
-- auxilary method and pattern matching and an accumulator
{-
rle [] = []
rle (h:t) = rle_acc t h 1

rle_acc [] b c = [(b,c)]
rle_acc (a:t) b c | a == b = rle_acc t b (c+1)
rle_acc (a:t) b c = [(b,c)] ++ rle_acc t a 1
-}

-- auxilary method and pattern matching and two accumulators, and tail recursion
{-
rle [] = []
rle (h:t) = rle_acc t h 1 []

rle_acc [] b c acc = acc ++ [(b,c)]
rle_acc (a:t) b c acc | a == b = rle_acc t b (c+1) acc
rle_acc (a:t) b c acc = rle_acc t a 1 (acc ++ [(b,c)])
-}

-- using foldl
{-
rle list = reverse $ foldl (rle_acc) [] list

rle_acc [] b = [(b,1)]
rle_acc ((last, count):rest) b | b == last = ((b, count+1):rest)
rle_acc acc b = ((b, 1):acc)
-}

-- without an accumulator
rle (a:b:rest) | a == b = let (_, count):rest2 = rle (b:rest)
                             in (b,count+1):rest2
rle (a:rest) = (a,1):rle rest
rle [] = []

main = print $ rle [1,1,2,2,2,4,3,3]

Stop Animation
2010-04-24T23:03:00.001-07:00
I keep looking up how to create an animated gif using ImageMagick and then forgetting, so here it is:convert mypref_??.jpg myanim.gifHere's one made from photos I took a while ago:

Were You Born To Be Agile?
2010-04-22T18:56:00.000-07:00
Bas Vodde put together a presentation that to me suggests that Americans are going to be culturally less compatible with Agile than Scandinavians:

BitTorrent URL BootStrap Idea
2010-04-17T08:42:00.000-07:00
So finally Day[9] has found a host site, Blip.TV, which allows the users to download the video rather than watch it online and risk playback issues. Great. So what's the problem? It still takes a fair while to download 200Mb+ at a max of 100K/s.

Here's my idea:

BitTorrent's could be seeded by an HTTP URL. That way when downloading users can supplement the limited bandwidth of the host with portions that other users had downloaded. This isn't a totally new idea, Blizzard's download tool bootstraps in this way. Random access to the HTTP content is also possible through the use of the HTTP Content-Range header field, so it appears to be workable without any custom server-side support.

Perhaps this is something that could happily be turned into a browser plug-in? It could work something like this:

1. The plugin detects that a download has started and that the file size is over some limit, say 100Mb.
2. It then looks for a dht hash of the URL from an indexing service.
3. If a dht exists then it begins searching for other peers.
4. If no dht exists then once the download completes it generates a hash for the content and submits it to the indexing service.

In either case there would be some sort of seeding necessary by the plug-in. The main hassle would probably be working with or overriding the browsers default download strategy.

Great Use of Groovy ASTs
2010-03-28T11:21:00.001-07:00
Take a look at gcontracts. Here's their example usage:
@Invariant({ firstName != null && lastName != null })
class Person {
  String firstName
  String lastName
  String name

  public Person() {}

  public Person(final String firstName, final String lastName) {
    this.firstName = firstName
    this.lastName = lastName
  }

  @Override
  @Requires({ firstName.size() > 0 && lastName.size() > 0 })
  def String toString() {
    return "${firstName} und ${lastName}"
  }

  @Requires({ delimiter in ['.', ',', ':'] })
  @Ensures({ name.size() > 1 } )
  def initCompleteName(String delimiter) {
    name = "${firstName}${delimiter}${lastName}"
  }
}Requires are the preconditions and Ensures are the post-conditions.

StarCraft2 AI Coding
2010-03-19T12:39:00.000-07:00
So I was curious as to how come the StarCrack AI folks are finding it so easy to develop new and improved AI for StarCraft 2. It turns out that there's an SC2Data file that's an mpq archive. So I downloaded their AI distribution and extracted their code. The first thing I noticed is that there's more than AI in here, there's actual game balancing information. E.g.:
<CWeaponLegacy id="Marauder">
  <EditorCategories value="Race:Terran"/>
  <DisplayEffect value="MarauderU"/>
  <DisplayAttackCount value="1"/>
  <TargetFilters value="Ground,Visible;Missile,Stasis,Dead,Hidden,Invulnerable"/>
  <Range value="6"/>
  <Period value="1.5"/>
  <DamagePoint value="0"/>
  <Backswing value="0"/>
  <Effect value="MarauderLM"/>
  <Icon value="Assets\Textures\btn-upgrade-terran-infantryweaponslevel1.dds"/>
</CWeaponLegacy>That makes me wonder if it's possible to edit these files in order to artificially make an opposing race harder or easier to play against.

Looking at the contents of TriggerLibs, this all looks very much like C code. Sweet. My next thought is that this can probably be translated into one big Groovy script without too much hassle, and then these StarCrack folks could write test harnesses! It must be a real pain to have to crank up StarCraft itself to see if you coded the right thing. I don't see all the TriggerDebugOutput logging as being a good substitute.

Edit: I gave this a quick stab. The first real problem with this idea is that there are multiple declarations of the method InitCounters.

Where Does the Time Go?
2010-02-28T19:29:00.000-08:00
Here, here, here, and here. With all that, who's got time for technology and blogging? ...Seriously.

Mvn Voodoo - eclipse:eclipse Doesn't Include AspectJ in the Classpath
2010-01-20T16:55:00.000-08:00
Nice huh? Well, the answer is in MECLIPSE-544. I added this to my pom.xml to fix the problem:<build>
  <plugins>
    <plugin>
      <groupId>org.apache.maven.plugins
      <artifactId>maven-eclipse-plugin
      <configuration>
        <!-- http://jira.codehaus.org/browse/MECLIPSE-544 -->
        <ajdtVersion>none
      </configuration>
    </plugin>
  </plugins>
</build>

Upgrade Complete
2010-01-17T21:56:00.001-08:00
if you've played any online flash games you may well find this amusing:

[Upgrade Complete]

Test & Code Edit Patterns - Part III
2010-01-16T17:52:00.000-08:00
Okay. It works. Here's a picture with the Arduino showing that my main code is 8 minutes older than my test code. 0-2 minutes: one LED, 2-4 minutes: two LEDs, 4-8 minutes three LEDs, 8-15 minutes: four LEDs, 15+ minutes: four flashing LEDs:

Here's my Arduino code for driving the display:#define RED 10
#define GREEN 11

void setup() {
  pinMode(RED, OUTPUT);
  pinMode(GREEN, OUTPUT);

  Serial.begin(9600);
}

int code = 0;
int test = 0;

void loop() {
  if (Serial.available() > 0) {
    int incomingByte = Serial.read();

    code = incomingByte / 16;
    test = incomingByte % 16;
  }

  draw(code, 2);
  draw(test, 8);

  if (code == 15 || test == 15) {
    delay(100);
    if (code == 15) {
      erase(2);
    }
    if (test == 15) {
      erase(8);
    }    
    delay(900);
  }
}

void draw(int value, int base) {
  digitalWrite(base+0, (value >= 1) ? HIGH : LOW);
  digitalWrite(base+1, (value >= 2) ? HIGH : LOW);
  digitalWrite(base+2, (value >= 4) ? HIGH : LOW);
  digitalWrite(base+3, (value >= 8) ? HIGH : LOW);
}

void erase(int base) {
  digitalWrite(base+0, LOW);
  digitalWrite(base+1, LOW);
  digitalWrite(base+2, LOW);
  digitalWrite(base+3, LOW);
}Here's my sloppy Groovy code for providing the data:import gnu.io.CommPortIdentifier
import gnu.io.SerialPort

CommPortIdentifier portId =
  CommPortIdentifier.getPortIdentifier(
    "/dev/tty.usbserial-A9007OMY")

SerialPort serialPort = (SerialPort) portId.open("Test", 5000) 

serialPort.setSerialPortParams(9600, 
  SerialPort.DATABITS_8, 
  SerialPort.STOPBITS_1, 
  SerialPort.PARITY_NONE) 

serialPort.setFlowControlMode(SerialPort.FLOWCONTROL_NONE) 

// give the Arduino time to reset and Serial.begin
Thread.sleep(3000) 

OutputStream outStream = serialPort.getOutputStream() 

walk = { root ->
  root.eachFile { file ->
    if (file.name.endsWith(".java")) {
      if (file.name.endsWith("Test.java")) {
        if (test == null || file.lastModified() > test.lastModified()) {
          test = file
        }
      } else {
        if (code == null || file.lastModified() > code.lastModified()) {
          code = file
        }
      }
    }
  }
  root.eachDir { dir ->
    if (!dir.name.startsWith(".")) {
      walk(dir)
    }
  }
}

while(true) {
  newestTest = 0
  test = null
  newestCode = 0
  code = null

  walk(new File(""))

  def now = System.currentTimeMillis()
  def codeAge = (now - code.lastModified())/60000
  def testAge = (now - test.lastModified())/60000

  println "Test (minutes old): $testAge ($test)"
  println "Code (minutes old): $codeAge ($code)"
  new File("monitor.csv").
    append("$now, $codeAge, $testAge\r\n")

  int encodedTest = (int)Math.max(Math.min(15,testAge-codeAge), 0)
  int encodedCode = (int)Math.max(Math.min(15,codeAge-testAge), 0)

  if (encodedTest == 0 && encodedCode == 0) {
    if (codeAge > testAge) {
      encodedCode = 1
    } else {
      encodedTest = 1
    }
  }

  int encoded = encodedCode * 16 + encodedTest

  outStream.write((byte)encoded)
  outStream.flush()

  Thread.sleep(20000);
}

outStream.close()
serialPort.close()

Test & Code Edit Patterns - Part II
2010-01-16T16:54:00.001-08:00
I'm having some success with this project. I can communicate with the Arduino using RXTX and Serial.read(). I'm deriving my last-edit times and encoding them as a byte and sending it over the USB to the Arduino. It can read the byte and update my row of LEDs accordingly.

But.

This only works for some reason if I have the Serial Monitor window open in the Arduino IDE. There must be some communication going on with the Arduino that puts it in a happy state for receiving data. The only hint is that the LED marked L flashes when I open and also when I close the Serial Monitor window. Something is going on there...

Do you know the answer to my problem? I'm downloading the Arduino IDE code now to see if I can find out what's going on.

Edit: I've been using this guide. From looking at the source code app/src/processing/app/Serial.java I'm using the same API and constructing it similarly:

Their code:port = (SerialPort)portId.open("serial madness", 2000);
input = port.getInputStream();
output = port.getOutputStream();
port.setSerialPortParams(rate, databits, stopbits, parity);
port.addEventListener(this);
port.notifyOnDataAvailable(true);My code:SerialPort serialPort = (SerialPort) portId.open("Test", 5000) 

serialPort.setSerialPortParams(9600, 
  SerialPort.DATABITS_8, 
  SerialPort.STOPBITS_1, 
  SerialPort.PARITY_NONE) 

serialPort.setFlowControlMode(SerialPort.FLOWCONTROL_NONE) 

OutputStream outStream = serialPort.getOutputStream()I've found no documentation for the CommPortIdentifier.open method - just this javadoc.

I'm going to try mimicking the Arduino IDE code more closely...

Edit: I think I have the solution!

Opening communication with the Arduino resets it! Serial.begin is the first thing my Arduino code does in it's setup method and that seems to need some time to settle down before anything can be read using Serial.read. Furthermore, closing communication with the Arduino resets the damn thing again! I hope this is configurable somewhere...

Edit: Humph. It's by design. You can find out more about it here.

Test & Code Edit Patterns
2010-01-15T08:24:00.000-08:00
I put together a Groovy script to monitor my coding pattern of behaviour. The idea is that I can monitor when I last changed a source file or a test source file. Initially I plan to record a few days of data and graph it. My grand plan is to use my Arduino to drive some LEDs to tell me how long it's been and indicate when I've spent to long coding, writing tests, etc.newestTest = 0
newestCode = 0

def walk(root) {
  root.eachFile { file ->
    if (file.name.endsWith(".java")) {
      if (file.name.endsWith("Test.java")) {
        newestTest = Math.max(newestTest, file.lastModified())
      } else {
        newestCode = Math.max(newestCode, file.lastModified())
      }
    }
  }
  root.eachDir { dir ->
    if (!dir.name.startsWith(".")) {
      walk(dir)
    }
  }
}

walk(new File(""))

def now = System.currentTimeMillis()
def code = (now - newestCode)/60000
def test = (now - newestTest)/60000

println "Test (minutes old): $test"
println "Code (minutes old): $code"