I agree with you.... Blueprint and CDI has this advantage over DS.
But I think I never had the necessity to inject beans in many years of development with OSGi. Injecting simple components and services fitted well all my needs. :) And the best from DS is that it is smart, small and fast.
As Blueprint, CDI and DS uses xml we can provide all. But I don't know If there is a way to choose only one at runtime... I mean, what could happen in a system where we have both Blueprint and DS active?2013/3/28 Charles Moulliard <ch007m@gmail.com>There is nevertheless one advantage of Aries Blueprint over Declarative Service which is :"Unlike Declarative Services, Blueprint can inject not only OSGi services, but also managed beans, i.e. plain old Java objects. Managed beans are local to the current component and cannot be injected into other components. Each Blueprint-enabled bundle has its own Blueprint context with managed beans and OSGi service references. The contexts of different bundles can interact by injecting services published by another bundle."On Wed, Mar 27, 2013 at 8:26 PM, Charles Moulliard <ch007m@gmail.com> wrote:
Great !On Wed, Mar 27, 2013 at 7:04 PM, Cristiano Gavião <cvgaviao@gmail.com> wrote:
Hi Charles,
Step one (installation of bundles and its dependencies) seems to be ok ;)
[INFO] drools-osgi-enviroment-tests ...................... SUCCESS [6.559s]
[INFO] drools-osgi-enviroment-tests-common ............... SUCCESS [15.887s]
[INFO] drools-osgi-enviroment-tests-equinox-juno ......... SUCCESS [20.176s]
[INFO] drools-osgi-enviroment-tests-equinox-kepler ....... SUCCESS [17.581s]
[INFO] drools-osgi-enviroment-tests-felix ................ SUCCESS [15.874s]
[INFO] drools-osgi-enviroment-tests-jbosgi ............... SUCCESS [39.769s]
[INFO] drools-osgi-enviroment-tests-knoplerfish .......... SUCCESS [14.839s]
[INFO] ------------------------------------------------------------------------
next step is to reuse the existent tests-jars
cheers,
Cristiano2013/3/27 Charles Moulliard <ch007m@gmail.com>
I think that those packages have been added since JDK 1.7 xxx. Reason why you should deploy the following bundles :with these versionsand more specifically 2.2.1 for jaxb-xOn Wed, Mar 27, 2013 at 4:32 PM, Cristiano Gavião <cvgaviao@gmail.com> wrote:
The packages that I'm aware of until now is:
- org.apache.poi.openxml4j.exceptions
- org.apache.poi.ss.usermodel (note that poi.jar exports this too)
I'm getting another problem with servicemix bundles related to com.sun.tools.xjc. it seems that there is no bundle exporting com.sun.source.tree that is being required:
ERROR: Bundle org.drools.decisiontables [38] Error starting mvn:org.drools/drools-decisiontables/6.0.0-SNAPSHOT (org.osgi.framework.BundleException: Unresolved constraint in bundle org.drools.decisiontables [38]: Unable to resolve 38.0: missing requirement [38.0] osgi.wiring.package; (osgi.wiring.package=org.drools.compiler.compiler) [caused by: Unable to resolve 29.0: missing requirement [29.0] osgi.wiring.package; (osgi.wiring.package=org.drools.core) [caused by: Unable to resolve 30.0: missing requirement [30.0] osgi.wiring.package; (osgi.wiring.package=com.sun.tools.xjc) [caused by: Unable to resolve 32.0: missing requirement [32.0] osgi.wiring.package; (osgi.wiring.package=com.sun.source.tree)]]])
org.osgi.framework.BundleException: Unresolved constraint in bundle org.drools.decisiontables [38]: Unable to resolve 38.0: missing requirement [38.0] osgi.wiring.package; (osgi.wiring.package=org.drools.compiler.compiler) [caused by: Unable to resolve 29.0: missing requirement [29.0] osgi.wiring.package; (osgi.wiring.package=org.drools.core) [caused by: Unable to resolve 30.0: missing requirement [30.0] osgi.wiring.package; (osgi.wiring.package=com.sun.tools.xjc) [caused by: Unable to resolve 32.0: missing requirement [32.0] osgi.wiring.package; (osgi.wiring.package=com.sun.source.tree)]]]
I'm using this deps:
<dependency>
<groupId>org.apache.servicemix.bundles</groupId>
<artifactId>org.apache.servicemix.bundles.jaxb-xjc</artifactId>
<version>2.2.6_1</version>
</dependency>
<dependency>
<groupId>org.apache.servicemix.bundles</groupId>
<artifactId>org.apache.servicemix.bundles.jaxb-impl</artifactId>
<version>2.2.6_1</version>
</dependency>
<dependency>
<groupId>org.apache.servicemix.specs</groupId>
<artifactId>org.apache.servicemix.specs.jaxb-api-2.2</artifactId>
<version>2.2.0</version>
</dependency>
<dependency>
<groupId>org.apache.servicemix.specs</groupId>
<artifactId>org.apache.servicemix.specs.activation-api-1.1</artifactId>
<version>2.2.0</version>
</dependency>
<dependency>
<groupId>org.apache.servicemix.specs</groupId>
<artifactId>org.apache.servicemix.specs.stax-api-1.2</artifactId>
<version>2.2.0</version>
</dependency>2013/3/27 Charles Moulliard <ch007m@gmail.com>
Can you please provide me the list of packages to be exported and I will include also poi-ooxml ?On Wed, Mar 27, 2013 at 3:03 PM, Cristiano Gavião <cvgaviao@gmail.com> wrote:
2013/3/27 Charles Moulliard <ch007m@gmail.com>
1) For Karaf project, we mostly use activator or Aries Blueprint (http://aries.apache.org/modules/blueprint.html). Declarative Service is rather new top of karaf (http://sully6768.blogspot.be/2012/09/scr-components-with-karaf.html)
As a consumer you can still using both blueprint or BundleContext.getService() to reference the services.
What I found good in both Blueprint and DS is that they ensure the component/services lifecycle and we could still use Configuration Admin with them.
The problem with blueprint is that is not default installed in most distributions while DS is.
Is DS installed by default in Karaf?
2) For poi-ooxml, we should use the ServiceMix bundle (http://repo1.maven.org/maven2/org/apache/servicemix/bundles/org.apache.servicemix.bundles.poi/3.9_1/) until they provide a OSGI bundle of POI.
Unfortunately this bundle don't export the package needed. it wraps the poi.jar but not poi-ooxml. org.apache.poi.ss.usermodel.Workbook.class :(
On Wed, Mar 27, 2013 at 1:53 PM, Cristiano Gavião <cvgaviao@gmail.com> wrote:
I won't change that code for while... It is still be there until you decide to move it to the new one. ;)
Btw, I have two questions:
1)what do you think about to use Declarative Services to register the services instead do it manually at activator ?
2) how are you dealing with poi-ooxml dependency in drools-decisiontables? there is no osgified jar for it...2013/3/27 Charles Moulliard <ch007m@gmail.com>
Hi Christiano,I prefer that first we finalize the OSGI bundles (drools-decisiontables, drools-jpa, drools-jbpm, ...) before changing pax-exam code which is working and also used by Apache Camel/Karaf projects ;-)Regards,CharlesOn Wed, Mar 27, 2013 at 1:44 PM, Cristiano Gavião <cvgaviao@gmail.com> wrote:
Hello,
I already look at it. The problem is that this tests are using an old version (2.x) of pax-exam and using a karaf specific api too.
That version will be dropped by karaf team soon. see [1] and [2].
Btw, Pax-exam was improved a lot in version 3.x.
[1] - http://karaf.922171.n3.nabble.com/Discuss-Karaf-and-Pax-Exam-3-x-td4028074.html
[2] - https://ops4j1.jira.com/browse/PAXEXAM-503
regards,
Cristiano2013/3/27 Charles Moulliard <ch007m@gmail.com>
Christiano,Can you please have a look here as there is already a pax-exam test for karaf (https://github.com/droolsjbpm/droolsjbpm-integration/tree/master/drools-osgi/drools-karaf-itest) ?Regards,CharlesOn Tue, Mar 26, 2013 at 11:11 AM, Cristiano Gavião <cvgaviao@gmail.com> wrote:
I'm creating a pax-exam project where I will run some test on top of equinox and felix.
After I commit and push it maybe you could add karaf stuffs. so it could help us to identify the reason of the error you are talking about...
On 25/03/13 14:19, Charles Moulliard wrote:
There was another error when using singleton := true.
On Mon, Mar 25, 2013 at 6:15 PM, Cristiano Gavião <cvgaviao@gmail.com> wrote:
well, I never seen any error related to singleton attribute at Felix or Equinox.
The error you have reported seems to be related to what is being done (and not being undone) inside the activator...
2013/3/25 Charles Moulliard <ch007m@gmail.com>
Not at all but using singleton := true option generates error when we do a stop, update start on Apache Karaf.We never used that property to generate all the bundles that we have in the project Karaf, ServiceMix, Geronimo, ...
On Mon, Mar 25, 2013 at 4:34 PM, Cristiano Gavião <cvgaviao@gmail.com> wrote:
Charles,
I saw that you removed singleton:=true in your commit. This is because do you plan to have more than one version of drools/jbpm running at same time?
regards,
Cristiano--
2013/3/25 Cristiano Gavião <cvgaviao@gmail.com>
Charles,
I already changed the manifest generation of kie and other drools modules. I created some pull requests for such changes.
Next thing that I planned to do this week(Wed) was to review each Activator, I think we could improve it...
regards,
Cristiano
On 25/03/13 04:52, Charles Moulliard wrote:
Hi,
The org.kie.api.osgi.Activator class of kie project needs the class ServiceRegistryImpl (& Interface ServiceRegistry of kie internal) to register an OSGI Service (Interface)
this.serviceRegistry = bc.registerService( ServiceRegistry.class.getName(),ServiceRegistryImpl.getInstance(),new Hashtable() );
but the maven module kie api does not have a dependency with kie internal as the class ServiceRegistry & ServiceRegistryImpl are part of the module kie internal & package org.kie.internal.utils
Questions :- What are the plans regarding to the Activator of Kie api bundle - what does it want to do ?- Can we add the missing dependency in kie api project ?
Regards,--
Charles MoulliardApache Committer / Sr. Enterprise Architect (RedHat)Twitter : @cmoulliard | Blog : http://cmoulliard.blogspot.com
_______________________________________________ rules-dev mailing list rules-dev@lists.jboss.org https://lists.jboss.org/mailman/listinfo/rules-dev
"Tudo vale a pena se a alma não é pequena..."
_______________________________________________
rules-dev mailing list
rules-dev@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/rules-dev
--
Charles MoulliardApache Committer / Sr. Enterprise Architect (RedHat)Twitter : @cmoulliard | Blog : http://cmoulliard.blogspot.com
_______________________________________________
rules-dev mailing list
rules-dev@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/rules-dev
_______________________________________________
rules-dev mailing list
rules-dev@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/rules-dev
--
Charles MoulliardApache Committer / Sr. Enterprise Architect (RedHat)Twitter : @cmoulliard | Blog : http://cmoulliard.blogspot.com
_______________________________________________ rules-dev mailing list rules-dev@lists.jboss.org https://lists.jboss.org/mailman/listinfo/rules-dev
_______________________________________________
rules-dev mailing list
rules-dev@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/rules-dev--Charles MoulliardApache Committer / Sr. Enterprise Architect (RedHat)Twitter : @cmoulliard | Blog : http://cmoulliard.blogspot.com
_______________________________________________
rules-dev mailing list
rules-dev@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/rules-dev
--
"Tudo vale a pena se a alma não é pequena..."
_______________________________________________
rules-dev mailing list
rules-dev@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/rules-dev--Charles MoulliardApache Committer / Sr. Enterprise Architect (RedHat)Twitter : @cmoulliard | Blog : http://cmoulliard.blogspot.com
_______________________________________________
rules-dev mailing list
rules-dev@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/rules-dev
--
"Tudo vale a pena se a alma não é pequena..."
_______________________________________________
rules-dev mailing list
rules-dev@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/rules-dev--Charles MoulliardApache Committer / Sr. Enterprise Architect (RedHat)Twitter : @cmoulliard | Blog : http://cmoulliard.blogspot.com
_______________________________________________
rules-dev mailing list
rules-dev@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/rules-dev
--
"Tudo vale a pena se a alma não é pequena..."
_______________________________________________
rules-dev mailing list
rules-dev@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/rules-dev--Charles MoulliardApache Committer / Sr. Enterprise Architect (RedHat)Twitter : @cmoulliard | Blog : http://cmoulliard.blogspot.com
_______________________________________________
rules-dev mailing list
rules-dev@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/rules-dev
--
"Tudo vale a pena se a alma não é pequena..."
_______________________________________________
rules-dev mailing list
rules-dev@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/rules-dev--Charles MoulliardApache Committer / Sr. Enterprise Architect (RedHat)Twitter : @cmoulliard | Blog : http://cmoulliard.blogspot.com
_______________________________________________
rules-dev mailing list
rules-dev@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/rules-dev
--
"Tudo vale a pena se a alma não é pequena..."
_______________________________________________
rules-dev mailing list
rules-dev@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/rules-dev--Charles MoulliardApache Committer / Sr. Enterprise Architect (RedHat)Twitter : @cmoulliard | Blog : http://cmoulliard.blogspot.com--Charles MoulliardApache Committer / Sr. Enterprise Architect (RedHat)Twitter : @cmoulliard | Blog : http://cmoulliard.blogspot.com
_______________________________________________
rules-dev mailing list
rules-dev@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/rules-dev
--
"Tudo vale a pena se a alma não é pequena..."
_______________________________________________
rules-dev mailing list
rules-dev@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/rules-dev