On Thu, Dec 3, 2015 at 9:20 AM, Romain Gilles <[email protected]> wrote:
> Some times maven-bundle-plugin can have side effect on > split package but I hope Raul has already checked it. > Not sure which side effect you mean. But anyway, the maven-bundle-plugin has several goals. There is a goal to delegate packaging to the plugin by using the "bundle" packaging, but I have not used that. We just invoke the manifest goal of the plugin to generate the MANIFEST.MF, and then we package the manifest into the JAR. The only things I mention to Raul is to double check that the user class > loader (here Raul provide an OSGi aware user classloader) is used every > where in the serialization process: > - Binary marshaller (new) > - Optimized marshaller (old) Correct. It was my reading of the source code that the user-set classloader (Ignite#setClassLoader) was not used in some cases (IGNITE-1930). But during my tests where I used the default marshaller with 1.5.0 (Binary Marshaller – Romain sorry for the confusion in Skype), I faced no issues for basic use cases. However, creating unit tests for OSGi is not a simple task and that's why I'd like to release this out to the public, so that everyone who's interested in OSGi can give it a spin with their use cases. As I said in IGNITE-1930, the usage of classloaders inside Ignite is choppy and should be improved to ensure that the user-set classloader is always used if set. We can declare OSGi functionality as "preview" – that would make me, and probably others, comfortable. Regards, *Raúl Kripalani* PMC & Committer @ Apache Ignite, Apache Camel | Integration, Big Data and Messaging Engineer http://about.me/raulkripalani | http://www.linkedin.com/in/raulkripalani http://blog.raulkr.net | twitter: @raulvk
