Wait, don’t panic, this blog post provides an explanation of how to raise Hudson jobs from their ashes!

How to find the issue

On the main Hudson page, go to Manage Hudson –> System Log –> All Hudson logs

You will find this kind of error:

SEVERE: Failed Loading job XXX-JOBNAME-XXX
hudson.util.IOException2: /var/lib/hudson/jobs/XXX-JOBNAME-XXX/nextBuildNumber doesn't contain a number
	at hudson.model.Job.onLoad(Job.java:369)
	at hudson.model.AbstractProject.onLoad(AbstractProject.java:342)
	at hudson.model.BaseBuildableProject.onLoad(BaseBuildableProject.java:102)
	at hudson.model.Items.load(Items.java:117)
	at hudson.model.Hudson$13.run(Hudson.java:2368)
	at org.jvnet.hudson.reactor.TaskGraphBuilder$TaskImpl.run(TaskGraphBuilder.java:146)
	at org.jvnet.hudson.reactor.Reactor.runTask(Reactor.java:259)
	at hudson.model.Hudson$4.runTask(Hudson.java:698)
	at org.jvnet.hudson.reactor.Reactor$2.run(Reactor.java:187)
	at org.jvnet.hudson.reactor.Reactor$Node.run(Reactor.java:94)
	at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1110)
	at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:603)
	at java.lang.Thread.run(Thread.java:636)
Caused by: java.lang.NumberFormatException: For input string: ""
	at java.lang.NumberFormatException.forInputString(NumberFormatException.java:65)
	at java.lang.Integer.parseInt(Integer.java:493)
	at java.lang.Integer.parseInt(Integer.java:514)
	at hudson.model.Job.onLoad(Job.java:366)
	... 12 more

In fact, the file nextBuildNumber mentioned in the stack is empty.

How to restore jobs

It is really simple (but can be really tedious if you have a lot of missing jobs).

Put a number in nextBuildNumber files for each job mentioned in the logs.

Be sure to put a number higher than the last job run.

Potential improvements

It would be nice to have the job visible in the job list even if the nextBuildNumber file is corrupted.
Perhaps we could put a warning decorator on this job, and also put an arbitrary value in this file from the UI.

I suppose the files were corrupted because the jobs were running when the VM crashed. So, it seems that this file is empty when a job is running. Could we prevent it from being “empty” for too long, and so minimize the opportunity for corruption in the event of a crash?

Add your 2 cents!

To continue this discussion, come on over to the Eclipse forum and share your ideas. I opened a topic just for this

Note: you might also have issues with fingerprints, but it shouldn’t be a blocking point. See this topic on the Eclipse forum for more details.

I’m happy to announce that BonitaSoft - the company I work for - is a sponsor of the event which will be focusing on “tools and techniques to support the software development process“.

There will be a great line-up including:

• an Eclipse Foundation Director Ecosystem Europe
• several Eclipse project leaders
• several Eclipse committers
• a Google Anita Borg Recipient 2012 winner

Please check the official wiki page for the  line-up (to be confirmed) and get the exact list of speakers and subjects.

Personally, I will be talking about Tools and Methodologies to leverage GMF power using an iterative and generative approach.

The talk will be based on our three years experience in developing Bonita Open Solution Studio, an RCP application with two friendly and highly customized GMF Editors.

We will be looking at the tools and methodologies we currently use to continuously improve, maintain and test our GMF editors.

Even after three years, we are still able to bring innovation to these editors. As a proof of our success, we have been awarded a prize for the “Best Modeling Application” in the Eclipse Community Award 2011.

I will also talk about the future, which tools and methodologies we plan to use, change, improve and keep.

See you there!

Nota: if unfortunately you can’t come to EclipseDay Delft and you would like to attend my talk, you can vote for my submission at EclipseCon Europe.

This tutorial will be very helpful in the following cases:

• NoClassDefFoundError:  there are two different versions of the same class in your classpath.
• Integration with unknown huge products (not osgi-ed):  detect multiple libs with the same version, and so reduce memory footprint.

From here look at the attribute _war and you will find from which archive the class was loaded.



Do you have some special tips to use MAT?

Note: this was part of my EclipseCon Europe 2011 talk and one of my Eclipse Democamp Grenoble 2011 talk.

The issue

GMF is a graphical framework, so there are User Interface tests. Unfortunately,  they failed with the following error – which you can find in .log of the workspace/.metadata folder – when you run any test on the eclipse server:

The solution

I spent some time on Google and found the solution… in several chunks. So I put them together to provide the solution here in a single piece:

- Check Run Xvnc during build (and don’t bother to check take screenshot, it doesn’t work)

- Add an Execute shell build action before launching your tests with metacity –replace –sm-disable & :

Simple but it’s all in one place. Hope it will help you.

And – talking about help – any help on understanding why these three GMF tests are still failing is very welcome.

It was my second EclipseCon Europe. And the first difference that I encountered between the two conferences was during my trip

from Grenoble to Ludwigsburg. Last year, during the flight, the Swiss airline company offered me chocolates – this year the German airline gave me a camembert sandwich. (Which would you prefer?) OK, time to talk about #ECE2011 more seriously. I can’t cover all the great things so here is just a selection.

If it was just this…

My favorite talk was the blockbuster “Collaborative Modeling: The New Deal“. Effectively, there was some great stuff that really improves collaborative modeling. It covered all the way from real-time graphical collaborative modeling with Dawn, to better handling of conflict resolution for Graphical Modeling Framework, to task focused modeling with Mylyn bridge.

Tutorial dilemma

I faced a big dilemma: to choose which tutorial to attend. I would have loved to go to the Tycho or Git tutorial, but I ultimately decided to attend the Jubula Tutorial and I have no regrets. The tutorial was really well conducted. About the product itself, my impression is that Jubula is powerful but not intuitive. I would appreciate more usable User Interface.

Modeling Symposium

This year, the Modeling Symposium was hosted as a BoFs. There was a lot of good stuff presented. I’m happy about the announcement of the GMF-Tooling revival. If you want more details on the symposium, I invite you to see Mariot Chauvin‘s blogpost.

Keynote

I was very interested in the keynote about modeling by Bran Selic. He explained his vision of modeling through eyes that have already seen a lot of material on the subject. I admit that I agree with his vision that modeling is an evolution and a more and more useful concept when going to more and more complex systems.

My talk

This talk was in the top 10 most scheduled and was quite well attended. The main objective of my talk was to show that there are Eclipse tools that are easy to use to solve performance issues. So, I’m happy that some of the attendees came to me after the talk and said that they will try to use some tools that I presented. You can find a recording of the talk here (oops they cut off the beginning). Based on feedback, I proposed a talk for EclipseCon 2012 focused specifically on the Memory Analyzer Tool. Comments and votes for it are welcome

I’m looking forward to next year’s EclipseCon Europe and also hope to attend my first EclipseCon North America next year too.

BonitaSoft Presentations

I presented the two highly customized GMF-based editors of Bonita Open Solution Studio during the Modeling Symposium (see my presentation below).  The goal of the presentation was to show customization that can be done with GMF while keeping agility for development. You can find the slides here. I’m pretty happy with the results, as I got good feedback and people came to ask more details on how the BonitaSoft Team does this.

By the way, we plan to submit a more complete and technical talk for EclipseCon 2011, in which we will show more about how to “pimp a GMF editor” with concrete and practical tips. Please tell us if you want us to include some specific content.

We also presented a poster. It was interesting to meet potential end users of the product, and people that are interested in the Eclipse technologies we are using (and sometimes both at the same time).

People

ESE was a great opportunity to meet a lot of people, contributors, commiters, and end-users. I thank all of them for the excellent exchange of knowledge and know-how.

There were also a few good runners who were able to wake up early in the morning. This is a good way to start a conference day.

Talks

There were lots of great talks. I’ll highlight 2 of them, my favorites.

The first was “Scale, Share and Store your Models with CDO,” about the new features of CDO. There was also a demo of Dawn. This allows real collaborative work with GMF editors using CDO technology. Modifications on a diagram appear in real-time on the diagram of another instance of the editor, or on another instance of a web-based GMF-editor!! It also shows potential conflicts on the diagram in real-time. In a nutshell, this talk was really amazing.

The second is “Distilling Heap Dumps: a Guide to Using and Extending Memory Analyzer.” I was already aware of some basic usage of MAT, but there is so much more that I learned during the talk. For instance, you can use MAT to observe the Thread or Bundle state. MAT also provides extension points to add customized queries, customized reports, customized names for better rendering, and more.

Advanced Technologies Experimentation

There were more than a few great experts at the conference. We took advantage of that and did some “advanced” experimentation.

With the help of Kai Töedter and his blogpost, Mickael set up the CSS styling of e4 in Bonita Studio, our RCP application based on Eclipse 3.6.

Mickaël also began to improve the build of SWTBot project by using Tycho. You can follow it via the mailing list.

On my side, I pair-programmed with Mariot Chauvin in order to improve support of SVG figures in the GMF editor. I spotted a memory issue some time ago (using MAT, btw) and this was a great opportunity to look at it deeper.

We plan to write some blogposts soon about this stuff. So stay tuned

And in conclusion…

In short, this conference was great. We came back with a lot of contacts and new knowledge. We are also proud to have shared the expertise we have acquired while developing Bonita Studio.

One final piece of advice and personal reminder: don’t leave the paper wrapping of the SwissAir chocolate in your pocket – it might earn you an advanced security check!

1. Starting the processus requires only two clicks. In the menu, choose “File->Acquire Heap Dump…”
   
2. The wizard that will guide you prompts. You can see running java processes.
   
3. Click on Configure… A new dialog will be displayed. You can specify here which provider to use to make the heap dump.
   You will see that two providers are available by default:
   - jmap tools (for Hotspot VM for instance)
   - an helper IBM VM.
   
4. I’m using an Hotspot VM, so I need to specify only the JDK_HOME of my JDK. (This step is required only the first time.)
   
   Be careful: a JDK is required, a JRE is not enough. Are you really the end user that solves my memory issues?
5. Apply the configuration of the Heap Dump Provider and click OK
6. Now you’re back to the first dialog, select the process that you want to analyze.
   You will notice at the bottom that a path is automatically generated for the heap dump file. By the way I recommend to choose a folder containing only dumps. MAT generates a lot of other files aside it (for indexation purpose I assume). So if you don’t want that it pollutes your home folder (which is used by default), you know what you have to do
   
7. Click on Finish
8. Wait for the heap dump to be written to the disk and parsed by MAT. These operations can take some time for huge java heap size.
9. A new dialog prompts, for a quick overview, keep “Leak Suspects Report” checked.
   
10. Click Finish
11. All right, you have the Heap Dump opened in MAT, ready to be analyzed!

In effect, this way requires that the process is still running. It is a debug/development use case. You still need dumps on OutOfMemoryError for production environment. You can find some information on the way to do that on the MAT wiki.
To learn more things about MAT, I recommend you to follow the MAT webinar which happens on September 8th.

Dynamic Services

This one was the trickiest to identify and fix because the only way to find it is in a stacktrace with NPE in the log. In fact, FrameWorkAdminService is launched via Declarative Services in 3.6 and so you need to add org.eclipse.equinox.ds in your plugin-based product definition. And be careful, this plugin is not identified as required. This is the “drawback of using Dynamic Services” (to quote Lars Vogel). This same issue is also problematic with e4. You can find the related bug here.

GMF

The issues

• In order to identify potential issues we tested the migration with RC1 and RC2. There were some API modifications that were out of phase between GMF and qvtOML with the RC1 and RC2 releases. It took some time to find them and we needed the qvtOML milestones update site to make it work.
• On one of our high custom editors, we could no longer reparent (move a figure inside another). Canonical Edit Policies has been improved in order to fix this annoying bug. For that reason, a new method has appeared in GMF templates to override the refreshSemantic method. So, first, I removed the overridden method — but the copy/paste didn’t work after that. The trick is to keep the overridden method and add to the end of this method, not at the beginning, a call to the super method. You can take a look at this bug and see our aspect template for more information.

The advantage

Using providers and Aspect templates limits the amount of customized code to check. In fact most of the modifications were done to remove some fixes as they are now in the new release (canonical diagram empty after restart, save as saving only diagram and others). So moving to Helios simplifies Bonita Open Solution code.

Tests

There is very good news here: Eclipse is now using JUnit 4 by default! Now you can write your plugin tests with JUnit4, and this avoids a lot of struggle trying to get both standard plugins tests and JUnit4/SWTBot tests working together. Both kinds of test can now run without any hack, using the provided JUnit, and the org.eclipse.ant.optional.junit fragment loaded in your product.

Target platform

• Kudos to greclipse which provides a zipped update site to test on Helios. However upgrading the target platform for greclipse 2.0.2 for Eclipse 3.6 was a little complicated due to a patched org.eclipse.jdt.core (can’t uninstall on the target platform/conflict with patched jdt.core).
• It seems that BIRT doesn’t provide an update site to test 2.6 on RC releases. So we kept the 2.5.1 version of BIRT. The good news is that the 2.5.1 version works great with Eclipse 3.6.
• A little tip: use p2 to construct your target platform and keep all feature names with their repositories. They will be helpful for your next migration.

In a nutshell, there are difficulties migrating to Eclipse 3.6 but it is definitely do-able!

I wish you a good migration, hoping that my feedback will help you, and look forward to your tips and feedback from your successful migration.

You may already know it, but Bonita Open Solution embeds Jetty when it comes to deploying your process-driven application. I tried to programmatically hot deploy an application with Jetty a few days ago but unfortunately didn’t find any documentation on how to do this. After a lot of trial and error, I finally did it! I was mainly inspired from the depths of a mailing-list archive (BTW when will we have a Jetty forum?). Here is the results of my research…

I found several ways to use hot deployment with Jetty, including:

- Context deployer. But this requires XML file configuration and replacing a war in a directory… and then waiting for a scan of the directory!!

- OSGI service. This seems to be a great and easy solution but it also seems to be only available for Jetty 7.

And now it’s high time to show you how to use hot deployment programmatically:

Note: As Bonita Open Solution is based on Eclipse 3.5, we are using Jetty version 6.

1. Create the server:

   1	Server jetty = new Server();

2. Create a handlerCollection with a ContextHandlerCollection in it:

   1 2 3 4

   HandlerCollection hc = new HandlerCollection (); ContextHandlerCollection contextHandlerCollection = new ContextHandlerCollection(); hc.setHandlers(new Handler[]{contextHandlerCollection}); jetty.setHandler(hc);

3. Start the server:

   1	jetty.start();

4. Hot deploy your web application:

   1 2 3 4

   WebAppContext warWebappContext = new WebAppContext(); //configure it ... contextHandlerCollection.addHandler(warWebappContext); warWebappContext.start();

5. Hot undeploy your application:

   1 2 3 4 5 6

   //get the WebAppContext to undeploy... WebAppContext toUndeploy = getWebAppContextToUndeploy(); //stop it toUndeploy.stop(); //and remove it from the handler collection contextHandlerCollection.removeHandler(toUndeploy);

In a nutshell, the trick is to use a HandlerCollection for your WebAppContextHandler.

I hope that this article will help others to hot deploy with Jetty. You’re welcome to  suggest better ways or to provide links to good documentation on this subject!

One way to customize your GMF editor is to modify the XPand templates. GMF provides a convenient way to do it. This post will explain two ways to take advantage of this method. To illustrate, I will show you how to set a custom SelectionManager in the editor.

Import GMF templates and customize them

1. First, import the base templates used for GMF generation. You can get them from the templates folder in org.eclipse.gmf.codegen plugin. Open plug-in view, right-click on the plugin and import it as source project:
2. Now you can see the templates in your workspace:
3. Create a folder which will contain your customized templates:
4. Find the file that you want to customize. Be sure to use the same hierarchy for your customized templates folder. For instance, take the Editor.xpt file: and put it in the corresponding folder:
5. Customize the template by modifying the DEFINE block that you want. For instance, to add your own SelectionManager, modify the configureGraphicalViewer DEFINE block:

   1 2 3 4 5 6 7 8 9 10 11 12

   «DEFINE configureGraphicalViewer FOR gmfgen::GenEditorView-» «EXPAND xpt::Common::generatedMemberComment» protected void configureGraphicalViewer() { super.configureGraphicalViewer(); «EXPAND xpt::editor::DiagramEditorContextMenuProvider::qualifiedClassName FOR editorGen.diagram» provider = new «EXPAND xpt::editor::DiagramEditorContextMenuProvider::qualifiedClassName FOR editorGen.diagram»(this, getDiagramGraphicalViewer()); getDiagramGraphicalViewer().setContextMenu(provider); getSite().registerContextMenu(org.eclipse.gmf.runtime.diagram.ui.actions.ActionIds.DIAGRAM_EDITOR_CONTEXT_MENU, provider, getDiagramGraphicalViewer()); «REM»Disable multiple selection for PREVIEW «ENDREM» getDiagramGraphicalViewer().setSelectionManager(new org.bonitasoft.studio.common.diagram.tools.BonitaDiagramSelectionManager()); } «ENDDEFINE»

6. For the GMF generator to  use your custom templates, there are two attributes to set in the gmfgen file:
   - Set Dynamic Templates value to true
   - Set Template Directory value with the platform path to the folder which contains your customized templates:

And now for the elegant solution

Ok, now you know how to customize your templates. Let’s go further – learn how to customize them more efficiently. You can use aspects to do this. Aspects allow you to override just a part of the template. This avoids duplicate code, is easier to read, more efficient to maintain and easier to upgrade. Yes I, and the Bonitasoft R&D team, recommend this way ^^

1. In order to do this, create a folder named “aspects”. Put your customized template into it, again respecting the hierarchy:



2. The customization is not exactly the same. This time, pick up just the DEFINE block that interests you, replace “DEFINE” with “AROUND” and then modify the contents of this block as above. This way, you have only the modified part of the template in the file. This is much clearer :

   1 2 3 4 5 6 7 8 9 10 11 12

   «AROUND configureGraphicalViewer FOR gmfgen::GenEditorView-» «EXPAND xpt::Common::generatedMemberComment» protected void configureGraphicalViewer() { super.configureGraphicalViewer(); «EXPAND xpt::editor::DiagramEditorContextMenuProvider::qualifiedClassName FOR editorGen.diagram» provider = new «EXPAND xpt::editor::DiagramEditorContextMenuProvider::qualifiedClassName FOR editorGen.diagram»(this, getDiagramGraphicalViewer()); getDiagramGraphicalViewer().setContextMenu(provider); getSite().registerContextMenu(org.eclipse.gmf.runtime.diagram.ui.actions.ActionIds.DIAGRAM_EDITOR_CONTEXT_MENU, provider, getDiagramGraphicalViewer()); «REM»Disable multiple selection for PREVIEW «ENDREM» getDiagramGraphicalViewer().setSelectionManager(new org.bonitasoft.studio.common.diagram.tools.BonitaDiagramSelectionManager()); } «ENDAROUND»

   If you want more examples, take a look at the templates used for Bonita Open Solution Studio. They are available on the svn.
   And all your comments are welcome