Adding Bundle Symbolic Name instruction doesn't change Source bundle name about p2-maven-plugin

[tombujok]

In progress. It will be a bigger change, as I have noticed some other issues regarding the instructions

[kingargyle]

Not sure if I should report this here or open another bug, but it appears that the Bundle-SymbolicName doesn't always get set. In particularl I'm having issues with the jackson-core-lgpl.jar from maven which I need to change the symbolic name. I've specified it, but not being changed.

[nachtgold]

Maybe you can check in your changes until today in a separate branch and we could help you ?

[tombujok]

Thanks for your feedback. I have almost finished the implementation though. I will try to push it in the next days.

If you would like to contribute I will create some feature-requests issues so that you can work on that. Thanks!

[kingargyle]

Any update pn this, or a suggestion where this bug is. I'm running into it with the following as well.

                 <artifact>
                   <id>org.slf4j:slf4j-api:1.7.2</id>
                   <source>true</source>
                   <transitive>false</transitive>
                   <instructions>
                      <Bundle-SymbolicName>org.slf4j.api</Bundle-SymbolicName>
                      <Bundle-Version>1.7.2</Bundle-Version>
                   </instructions>
                </artifact>

I may have to go through and try to fix this locally, since I can create test case for this.

[tombujok]

Hi, sorry I did not have much time to finish this. I have fixed it along with some other bugs. I am working on it just now coding some integration tests. I hope I will manage to commit it today or tomorrow.

[tombujok]

Please recheck now. There's a lot that has been changed. This bug should be solved now.

• The override option is more intuitive.
• The handling of the instructions has been changed. Please see: https://github.com/reficio/p2-maven-plugin/blob/master/examples/quickstart/pom.xml
• The logging has been improved
• Tycho has been changed to 0.16.0 (and unified)
• I also added an option to exclude specified transitive dependencies (line in maven) using the switch. Please see: https://github.com/reficio/p2-maven-plugin/blob/master/examples/excludes/pom.xml
• Finally, an integration tests layer has been added. There's around 20 integration tests that invoke the plugin and verify the results.

[tombujok]

The newest snapshot should be build and deployed at night

[tombujok]

Snapshots were deployed to the reficio repo (http://repo.reficio.org/)

[kingargyle]

Thanks I'm testing this out now. Will report back the results.

[kingargyle]

There is a compilation error in the BundleWrapper code. I can submit a pull request to fix it but it is a one line change need to change: log.debug to log().debug. Once that is fixed, the renaming issue I was having seems to be fixed once I added the overrides.

[tombujok]

Are you sure you have the newest code?
The version from git compiles on jenkins (and on my machine) and there is no debug statement in the BundleWrapper code...
https://github.com/reficio/p2-maven-plugin/blob/master/src/main/java/org/reficio/p2/utils/BundleWrapper.java

[tombujok]

And the link to jenkins: https://reficio.ci.cloudbees.com/job/p2-maven-plugin/

[kingargyle]

Hmmm...odd line 137, used to contain the following line:

log.debug("Unsign Jar: " + jarToUnsign.getName());

Which was causing compilation errors. Latest code doesn't seem to contain that.

[tombujok]

OK. I understand - I will add this log line later.
The override option is required now - otherwise it may be difficult to spot mistakes if you specify instructions for artifacts that are already OSGi bundles.

So, may I close this issue?

[kingargyle]

Yep. I'm closing the issue.

[tombujok]

Cool, thanks. I will have a closer look at the code doing the code review, but I think we have got a release candidate for the 1.0.0 version.
I am pretty proud of the 19 integration tests (run by the maven-invoker-plugin) that increased the test coverage to around 90% :) Now, I am confident that the code is more or less stable...