First install ruby, gem and god.

apt-get install ruby, rubygems
gem install god

Then paste this initscript into /etc/init.d/god: got init script. Configure it to use /etc/god/config.rb for example, and there do some minor setup (using ruby, yay) so we can get emails if something happenes:

And then all that’s left to implement is a watch, so create an /etc/god/watches/xvbf.rb. It’s using a nice DSL to configure how God should act when a process dies etc:

The code is pretty much self explanatory. We watch if the process Xvfb is running, and if not, after the grace period we restart it. Optional info email could also be sent using the notify property.

That’s it, have fun automating your build environments!

Well, that was all nice and cool but git (other DVCS have this too) has a lot more to offer! :-) Today we’ll look into git bisect and how to automate it’s execution using maven to look for a commit that breakes something. You may run ANY SCRIPT and even write your own and git will use it to test if a commit is “bad” or “good”. Ok, but let’s slowdown a little… What git bisect basically does is performing an binary search, between a range of commits you specify looking for “the first bad commit” you you can easily find what caused the bug and fix it. You can drive it by hand and after starting git bisect just test the commit by hand and tell git if it’s good or bad by “git bisect bad” “git bisect good”. Git is smart enough to find what commit he should “git checkout …” next to perform a binary search based on your anwsers.

Today we’ll learn how to make git cooperate with maven and find the first commit that introduced a test failure. See bellow for the full workflow you showcasing this:

Ok up until now we’ve just created a simple repository we’ll be running the tests on. Let’s tag this state as “yeah it’s working here” just so that I don’t have to remeber the SHA-1 of this commit.

And now we’ll run a script that does some random commit action, just so that bisect has a lot of commits to work on.

The tests are still passing… time to change that! We’ll break the build with the following commit. And then generate yet another 50 commits to “hide it”… ;-)

Ok the test’s fail now. But suppose you didn’t know which commit introduced the bug (in our case made the test’s fail). You’d have to guess which one, or check them all or just work it out by looking at your current repository state. But since you’re doing small commits and do them often – it’ll be great if you found out which commit caused this failure. As it’s preferably small, it’ll be very easy to find out what change caused the build to fail. No is the time to use git bisect!

We’ve started bisect, and told him which commit we’re sure is working and which we’re sure is failing. Git will now checkout a commit he found “somewhat in the middle” and will continue to bisect (binary search) the commit history based on wether a commit is “good” or “bad”. We could just tell it git by hand by doing “git bisect good/bad” etc. But let’s git bisect run a script to do this for us :

Yeah, we’ve found our failing commit :-) You could use any script you want here, it just has to return 0 if git should think it’s a “good commit” and “not 0″ if it’s a “bad commit”. I’ve even seen runners written in PHP running or spellcheckers – you name it… :-) Ok but how does this magical script that we’ve just run look like?

Whoa, that’s simple… :-) Keep on git hacking guys! See you soon, with some more git or java tips… :-) Oh and I’ll post my presentation (holefully with video) from SFI soon, so keep an eye out! :-)

PS: Yup, the code retreat we’re (SCKRK with a little help from the PolishJUG) organizing got “sold out” after just 3h… ;-) Be sure to take a look and be quicker next time! http://coderetreat.sckrk.com

The code comes from http://arnorehn.de/cgi-bin/weblog_basic/index.php?p=34 so big kudos too him :-)

Cheers and happy hacking!

PS: I’m waiting for my T-shirt and stickers (for training participants) to arrive from github. I’d love they make it in time for SFI but I’m not sure if it’s even possible hmmm…

As I’d rather show the participants around git and not just “how the hell do I use vimdiff?!” I’m posting this little snippet on howto use KDiff3 on windows. Of course we’re probably going to have quite simple conflicts that could easily be conflicted without tooling but it’s nice to have one prepared when you need one. So, this is what you need to run KDiff3 on windows as a mergetool (paste this inside of your .git/config file)

[merge]
    tool = kdiff3

[mergetool "kdiff3"]
    path = C:/Program Files/KDiff3/kdiff3.exe
    keepBackup = false
    trustExitCode = false

And you’ll need to download kdiff3 for windows of course, this can be done on their sourceforga page. And that’s basically it. I’ll reconfirm that this really works tomorrow as I’m getting somewhat tired already… Well then…

Gundam fighters set, ready….. goooo!
* stupid quote from  G Gundam ;-)

UPDATE
The plugin is now available from Sonatype Nexus!

    
        sonatype-releases
        Sonatype Releases
        https://oss.sonatype.org/content/repositories/releases/
    

END OF UPDATE

Now I’ll be a bit lazy and just paste the README I’ve prepared for maven-git-commit-id-plugin on github.

  git clone git://github.com/ktoso/maven-git-commit-id-plugin.git maven-git-commit-id-plugin
  cd maven-git-commit-id-plugin
  mvn install

   <?xml version="1.0" encoding="UTF-8"?>
   <project xmlns="http://maven.apache.org/POM/4.0.0"
            xsi:schemaLocation="http://maven.apache.org/POM/4.0.0 http://maven.apache.org/maven-v4_0_0.xsd"
            xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">
       <modelVersion>4.0.0</modelVersion>

       <groupId>pl.project13.maven</groupId>
       <artifactId>my-git-plugin-sample-app</artifactId>
       <packaging>war</packaging>
       <version>0.1</version>
       <name>my-git-plugin-sample-app</name>
       <url>http://www.blog.project13.pl</url>

       <parent/>

       <dependencies />

       <build>
           <!-- GIT COMMIT ID PLUGIN CONFIGURATION -->
           <resources>
               <resource>
                   <directory>src/main/resources</directory>
                   <filtering>true</filtering>
                   <includes>
                       <include>**/*.properties</include>
                       <include>**/*.xml</include>
                   </includes>
               </resource>
           </resources>

           <plugins>
               <plugin>
                   <groupId>pl.project13.maven</groupId>
                   <artifactId>git-commit-id-plugin</artifactId>
                   <version>1.0-SNAPSHOT</version>
                   <executions>
                       <execution>
                           <goals>
                               <goal>revision</goal>
                           </goals>
                       </execution>
                   </executions>
                   <configuration>
                       <prefix>git</prefix> <!-- that's the default value -->
                       <dateFormat>dd.MM.yyyy '@' HH:mm:ss z</dateFormat> <!-- that's the default value -->
                       <verbose>true</verbose> <!-- false is default for this -->
                       <dotGitDirectory>${project.basedir}/../.git</dotGitDirectory> <!-- required, you have to specify this path -->
                   </configuration>
               </plugin>
               <!-- END OF GIT COMMIT ID PLUGIN CONFIGURATION -->

               <!-- other plugins -->
           </plugins>
       </build>
   </project>

 git.branch=${git.branch}

 git.build.user.name=${git.build.user.name}
 git.build.user.email=${git.build.user.email}
 git.build.time=${git.build.time}

 git.commit.id=${git.commit.id}
 git.commit.user.name=${git.commit.user.name}
 git.commit.user.email=${git.commit.user.email}
 git.commit.message.full=${git.commit.message.full}
 git.commit.message.short=${git.commit.message.short}
 git.commit.time=${git.commit.time}

<?xml version="1.0" encoding="UTF-8" standalone="no"?>
<beans xmlns="http://www.springframework.org/schema/beans"
       xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
       xsi:schemaLocation="http://www.springframework.org/schema/beans http://www.springframework.org/schema/beans/spring-beans-3.0.xsd">

    <bean name="gitRepositoryInformation">
        <property name="branch" value="${git.branch}"/>
        <property name="commitId" value="${git.commit.id}"/>
        <property name="commitTime" value="${git.commit.time}"/>
        <property name="buildUserName" value="${git.build.user.name}"/>
        <property name="buildUserEmail" value="${git.build.user.email}"/>
        <property name="commitMessageFull" value="${git.commit.message.full}"/>
        <property name="commitMessageShort" value="${git.commit.message.short}"/>
        <property name="commitUserName" value="${git.commit.user.name}"/>
        <property name="commitUserEmail" value="${git.commit.user.email}"/>
    </bean>

</beans>

package pl.project13.maven.example.git;

import org.codehaus.jackson.annotate.JsonWriteNullProperties;

/**
 * A spring controlled bean that will be injected
 * with properties about the repository state at build time.
 * This information is supplied by my plugin - <b>pl.project13.maven.git-commit-id-plugin</b>
 *
 * @author Konrad Malawski
 */
@JsonWriteNullProperties(true)
public class GitRepositoryState {
  String branch;                  // =${git.branch}
  String commitId;                // =${git.commit.id}
  String buildUserName;           // =${git.build.user.name}
  String buildUserEmail;          // =${git.build.user.email}
  String buildTime;               // =${git.build.time}
  String commitUserName;          // =${git.commit.user.name}
  String commitUserEmail;         // =${git.commit.user.email}
  String commitMessageFull;       // =${git.commit.message.full}
  String commitMessageShort;      // =${git.commit.message.short}
  String commitTime;              // =${git.commit.time}

  public GitRepositoryState() {
  }

  /* Generate setters and getters here */
}

<context:property-placeholder location="classpath:*.properties" />
<import resource="classpath:/git-bean.xml"/>

 @Controller
 @RequestMapping("/git")
 public class GitService extends BaseWebService {

     @Autowired
     GitRepositoryState gitRepoState;

     @RequestMapping("/status")
     public ModelAndView checkGitRevision() throws WebServiceAuthenticationException {
       ServerResponse<GitRepositoryState> response = new ServerResponse<GitRepositoryState>(gitRepoState);
       return createMAV(response);
     }
 }

 {
     "branch" : "testing-maven-git-plugin",
     "commitTime" : "06.01.1970 @ 16:16:26 CET",
     "commitId" : "787e39f61f99110e74deed68ab9093088d64b969",
     "commitUserName" : "Konrad Malawski",
     "commitUserEmail" : "konrad.malawski@java.pl",
     "commitMessageFull" : "releasing my fun plugin :-)
                            + fixed some typos
                            + cleaned up directory structure
                            + added license etc",
     "commitMessageShort" : "releasing my fun plugin :-)",
     "buildTime" : "06.01.1970 @ 16:17:53 CET",
     "buildUserName" : "Konrad Malawski",
     "buildUserEmail" : "konrad.malawski@java.pl"
 }

The G73 has an backlit keyboard, which may come in handy while late night coding without lights on as I like to code sometimes. (I’m using a X6 Sidewinder, even with this Notebook so it’s not like it’s keyboard matters to me most of the time ;-)) As usual with such “super cool addons” the funtion keys regulating the keyboard backlight seemed to only work on Windows – which I’m only using for gaming (but rarely) and testing if the software I happened to write runs ok on it. Anyways, turns out that the keyboard does really well integrate with linux, and you just need a few simple commands to use it, I’ve made them available on github :-)

https://github.com/ktoso/g73-keyboard-backlight-sh <- Go here to check it out :-)

git mv oldname newname
        =
mv oldname newname
git add newname
git rm oldname

Note that last line, git rm oldname is what we fotgot to do in our example, thus git was seeing an just the “new file” version of our operation. One line wrong, yet very important :-) Let’s check if it works this way:

[ktoso@homunculus test]$ ls
jeden  typescript
[ktoso@homunculus test]$ git init .
Reinitialized existing Git repository in /home/ktoso/test/.git/
[ktoso@homunculus test]$ cat jeden
drugi pliczek
[ktoso@homunculus test]$ echo 'dwadwa' > dwa
[ktoso@homunculus test]$ cat dwa
dwadwa
[ktoso@homunculus test]$ git add .
[ktoso@homunculus test]$ git commit -m 'dodanie ich'
[master (root-commit) a430782] dodanie ich
2 files changed, 2 insertions(+), 0 deletions(-)
create mode 100644 dwa
create mode 100644 jeden
create mode 100644 typescript
[ktoso@homunculus test]$ git mv
dwa         .git/       jeden       typescript
[ktoso@homunculus test]$ git mv jeden nowy-jeden
[ktoso@homunculus test]$ git status
# On branch master
# Changes to be committed:
#   (use "git reset HEAD <file>..." to unstage)
#
#       renamed:    jeden -> nowy-jeden
#
[ktoso@homunculus test]$ mv
dwa         .git/       nowy-jeden  typescript
[ktoso@homunculus test]$ mv dwa nowe-dwa
[ktoso@homunculus test]$ git add nowe-dwa
[ktoso@homunculus test]$ git rm dwa
rm 'dwa'
[ktoso@homunculus test]$ git status
# On branch master
# Changes to be committed:
#   (use "git reset HEAD <file>..." to unstage)
#
#       renamed:    dwa -> nowe-dwa
#       renamed:    jeden -> nowy-jeden
#
[ktoso@homunculus test]$ echo "problem Solved... :-)"
problem Solved… :-)

Yup, it actually detects the renames proppery. Even more, it would detect an “copy” action (this works even easier):

[ktoso@homunculus test]$ cp nowy-jeden nowy-jeden-jeden
[ktoso@homunculus test]$ git add .
[ktoso@homunculus test]$ git status
# On branch master
# Changes to be committed:
#   (use "git reset HEAD <file>..." to unstage)
#
#       renamed:    dwa -> nowe-dwa
#       copied:     jeden -> nowy-jeden
#       renamed:    jeden -> nowy-jeden-jeden
#       modified:   typescript
#

I’m already converting all the presentations to MP4 files and they’ll be online soon I hope :-) Then I’ll also write an usual “review” as I always do with any confernce/meering. For now just a quick summary: An amazing attandee count, nice presentations and an very very nice crowd :-) Thanks a lot for all the talks, but let’s wair for my “review” with more informations about JC5… :-) (Oh, yes we’re allready planing JC6… ;-))

As my blog is being listed on http://www.dworld.pl/Blogsfera (and for other obvious reasons) I didn’t want all my tweets to be translated “1 to 1″ to plain old blog posts since this would cause each of my tweets (which can be silly sometimes (but quite rarely!)) to land on dworld, which they obviously shouldn’t as it’s only an place for Java/Development oriented posts. Also I’d rather they’d take less space, and would skip all the categories/tags – it’s all in the tweet, so why repeat it again and again?

So… let’s start by importing tweets into wordpress. There’s an great plugin for that, and it’s called Tweet Import by skinju. Just install it and set it up to fetch your tweet’s at an regular basis. Ok, we’re done with step 1 – but neither are the current tweets smaller or ommitted in the rss feed. Let’s start by filtering them out from the rss feed:

Open up the file /wp-content/themes/default/functions.php and go to the end of the file, we’ll add an filter here. Please note that in step 1 you should have created an “twitter” category all tweets should be marked with. Let’s assume this category has an ID of 131, then the following code would do the filtering for us:

It’s a very nice trick, and it’s explained more in detail here. The code is really simple, so no explaining here and let’s go and tune our view (yup, the RSS feed is already fixed!).

To setup our view to work a little different for tweets than for posts, well have to edit: /wp-content/themes/maze/index.php so open it up in your favourite editor and add such an if statement inside an post’s while statement:

Ok, with this done let’s write the else part…

(You can complain about the in line css styling if you want later, I know it’s bad but this wordpress theme is a mess anyways… ;-)) Let’s have a look at the more important parts of this code part. You’ll notice that I’m using some weird get_post_meta() function. It’s an method that fetches additional information that each post CAN but doesn’t have to have. And as the TweetImport plugin does fill an post with a lot of useful tweet data, such as post time and tweet url – that’s what I’m fetching here. The rest should be quite self explanatory, there are bit’s of hardcoded “ktosopl” in there but hey – it’s just my blog and I’m a busy person ;-)

Hope you found this post interesting and maybe you’ll implement the same thing on your blogs…? :-)
Cheers!

PS: Oh, and if your interested in talking on one of your upcomming JavaCamps – feel free to contact me or anyone from the java.pl team, we’re looking forward to hearing from you. The nearest meeting will probably be organized around 23′th or 6′th…

Nuff said, let’s get down to some code. Basic android stuff is really easy to grasp and I wasn’t entirely “new” to it so coding went without big problems. One thing I missed was Dependency Injection – which I learned to love thanks to JEE – esp. Spring and Guice. But fear not – here’s how you can use DI – using Guice – in your android apps:

Just get the RoboGuice and Google Guice (the NO-AOP version) JARs and attach them to your project. Next, you’ll have to configure your “Application” in your AndroidManifest.xml:

The “android:name” specifies the “Application” class your app will use. This will be used by android as the “Application” instance, which would otherwise be just an plain old android “Aplication” – this is an common class for all your Activities and Services, you can do some global setup stuff here etc. And thats wat we are going to do right now:

We’re extending the GuiceApplication here, it’s provided by RoboGuice and allows us to easily add GuiceModules to our app – which then will be used in dependency resolution as they would normaly in any DI framework. If your interested, thats the ClassHierarhy we hooked into:

Ok, done. Now let’s setup Guice. It’s pretty straight foreward we bind interfaces to implementations Guice should use. If needed we could do some more interesting bindings (by our own annotations etc) but I don’t need such setups in my app so just such an module is ok:

And you’re ready to go. Now you can inject all you want in constructors/setters/fields, for example like this:

Please note that normally to obtain an SharedPreferences instance you would have to be in an Activity and get it like this:

Another nice thing RoboGuice allows us to do is injecting relations from our view.xml’s into our Java sources – which reminds me a little of GWTs UiBinder, but a little reversed, as in GWT the view has all the ui:field=”blaField” and here we use R.id.* to pass and get those references… The overall feel and look stays quite the same I guess – variables linked to XML objects in the layour. :-) Anyways, here’s how this linking looks like without DI:

And now the version using RoboGuice:

Much cleaner in my opinion, and it introduces less places we have to remember changing when our UI Object changes from an TextView ro an WebView for example… It’s also possible to inject Strings and other resources this way.

Oh and by the way, the project I took the code samples and learned all this stuff on is on my GitHub, here.

At work I’m working on an Ubuntu 10.4 (love you guys for getting me a PC with Linux without making a fuss about it!) my laptop is on Fedora 13 i586 and my home PC is running Fedora 13 x64. Thank goodnes they’re all linux… ;-) The problem is was that the Google Web Toolkit plugin isn’t working with Firefox 3.6.4 on an x64 system on linux! So… can’t I develop GWT stuff on my super powerful home PC? Of course I can, the only thing I needed to do, is to get the latest sources for the plugin and compile it for myself ;-) Here’s how:

And we’re done — off to GWT development!