Going through your extensions, and looking at the Advisor, I thought I'd mention that it is not a good idea to try to modify the work item in the Advisor - actually the extension point documents that this is not supported. It works, but you could get youself into serieous trouble in the future. See: http://rsjazz.wordpress.com/2012/12/14/do-not-modify-the-triggering-element-in-an-operationadvisor/

You should only modify work items in a follow-up action/participant.

Ralph,

Indeed the deployment works with your suggestions and I thank you for your help.
There was no problem with UTF-8 settings, as this is the default on Linux.

I think the explanation for the error is related to the component distinguished name.

Now I have another problem when the participant is invoked on workitem save. RTC throws this exception:

2013-04-24 23:53:05,163 [              WebContainer : 3] ERROR t.internal.registry.AbstractActivationManagerOwner  - The class "com.ibm.team.workitem.service.IWorkItemCommon" f
ailed to load. The bundle "itsm2.rtcext.workitem.extensions.service.participants" could not find the class.
java.lang.ClassNotFoundException: com.ibm.team.workitem.service.IWorkItemCommon
        at org.eclipse.osgi.internal.loader.BundleLoader.findClassInternal(BundleLoader.java:506)

This happens both on Tomcat and on Websphere Application Server.

concerning
com.ibm.team.workitem.service.IWorkItemCommon
not found

IWorkItemCommon class is in

com.ibm.team.workitem.common plugin
not
com.ibm.team.workitem.service plugin

The plugin is created on linux (Ubuntu) and I have JazzTeamServer deployed on Tomcat on the same linux while the WAS server is on an AIX machine.

However, as soon as I eliminate the "extends AbstractService" from the class definition, the plugin is loaded OK and I can see the component listed in the CCM's Admin ComponentStatus page. I can also see the advisor and the participant registered because I can associate them with the work item save operation.

In this scenario, as soon as I create a new workitem, fill data in and press the save button, I get the error about not being able to instantiate the advisor / participant for not being of the AbstractService type.

Oh, and I do not think the zip error is related to the plugin because in the web interface, the "Provision Status" page shows some content but stops short of my plugin:

CRJAZ0299I Installing bundle from the URL "file:////rational/IBM/JazzTeamServer/server/conf/ccm/sites/update-site/plugins/com.ibm.rational.rrdi.etl.rif2irifconverter_1.1.1.v20120416_2135.jar".
CRJAZ0299I Installing bundle from the URL "file:////rational/IBM/JazzTeamServer/server/conf/ccm/sites/update-site/plugins/com.ibm.rational.rpe.core.doors_1.1.2.v20111212_1225.jar".
CRJAZ0299I Installing bundle from the URL "file:////rational/IBM/JazzTeamServer/server/conf/ccm/sites/update-site/plugins/com.ibm.rational.rpe.core_1.1.2.v20120531_2156.jar".
CRJAZ0299I Installing bundle from the URL "file:////rational/IBM/JazzTeamServer/server/conf/ccm/sites/update-site/plugins/com.ibm

These are the last 4 entries in the log. And I know my plugin is yet to be processed because it is located in a different folder than "update-site" which is not listed in the log.

Do you define an activator? Can you please remove the definition if so?

Otherwise there is maybe something smelly with the update site. Hard to tell. I am not convinced I can debug this just in the forum.

Indeed I have an activator. I will proceed to remove it and will let you know of the results tomorrow morning.

I have tried with a default public constructor. I have tried with no constructor in my code to let the java compiler generate one for me.

All my attempts were met with a server shot in pieces as long as I declare my class to extend AbstractService.

By the way, in your lab exercise, the code does not declare a constructor, although the class declaration does contain the extends statement. So I can only assume that the java compiler is used to generate a default constructor. How does it work for you? (I must admit I have not tried to deploy the plugins that result from your code samples. I do believe that the best lessons can be learned when doing not when observing something that was prepared by an expert.)

So, what else can I try?

Make sure to organize imports and that you pick up the right AbstractService.

I have the right import, exactly as in your code.

Also, my code looks exactly like yours as far as class and method declarations are concerned.

There must be something else that I am missing.

One thing I noticed is that when the server dies, it typically throws an exception related to the loading of some com.ibm.team.repository.<something> class which is kind of strange because I think that classes from this package are loaded before plugins are loaded.

If it runs on Jetty, please be aware that in the test environment you have client as well as server plugins. So you can end up mixing client code into your server plugin. Once going to Tomcat, you won't have that and you'd fail to deploy. If you could provide information for the <something> one could tell.

See http://rsjazz.wordpress.com/2013/02/28/setting-up-rational-team-concert-for-api-development/
and
http://rsjazz.wordpress.com/2013/03/14/what-apis-are-available-for-rtc-and-what-can-you-extend/ especially Don’t Mix Client and Server API

I think you raised a very important issue of not mixing the APIs. How do I verify if I am doing the mixing or not?

By the way, I am not deploying on Jetty. Instead I deploy directly in Tomcat and use remote debugging for development testing.

My plugin code imports the following packages:

import com.ibm.team.process.common.IProcessArea;
import com.ibm.team.process.common.IProcessConfigurationElement;
import com.ibm.team.process.common.advice.AdvisableOperation;
import com.ibm.team.process.common.advice.IReportInfo;
import com.ibm.team.process.common.advice.runtime.IOperationParticipant;
import com.ibm.team.process.common.advice.runtime.IParticipantInfoCollector;
import com.ibm.team.repository.common.IAuditable;
import com.ibm.team.repository.common.IContributorHandle;
import com.ibm.team.repository.common.TeamRepositoryException;
import com.ibm.team.workitem.common.ISaveParameter;
import com.ibm.team.workitem.common.model.ISubscriptions;
import com.ibm.team.workitem.common.model.IWorkItem;
import com.ibm.team.repository.service.AbstractService;

and the plugin XML makes reference to:

com.ibm.team.process.common;bundle-version="1.2.1000"
com.ibm.team.repository.common;bundle-version="1.3.1000",
com.ibm.team.repository.service;bundle-version="1.2.1000",
com.ibm.team.workitem.common;bundle-version="3.1.0"

while the feature XML makes reference to :

<import feature="com.ibm.team.process.server.jfs.feature" version="1.1.1000.v20120518_2049"/>
<import feature="com.ibm.team.repository.server.jfs.feature" version="1.0.1000.v20120518_2049"/>
<import feature="com.ibm.team.workitem.server.rtc.feature" version="3.1.0.v20120601_0115"/>

The Zip exception is disturbing. Can you unzip the Jar files on your target platform? I think this is the core problem. You could test if it deploys on the platform you used for development.

As mentioned before I managed to kill a ccm server by providing a broken extension as well to the point where it would not start.

Without your plugin source I am out of options.

The last test, for which I provided the Tomcat server log was performed on the same platform on which the update-site was generated. And the jar files can be unzipped (I have checked that).

If you want, I can send you my source projects as well as the generated update site for you to take a look at.

Do that, no promises.

Understood. Thanks for your assistance. I have sent you the projects.