I have had success embedding transitive dependencies with the bundle goal and the Embed-Dependency/Embed-Transitive instructions.
My use case is to take a pom with a list of maven dependencies and create a directory that can be used as an OSGi "target platfom" in Eclipse. The downside of this approach (Embed-Dependency) is that it creates one really heavyweight uber bundle that defeats the modularity benefits of OSGi. Furthermore, it is a blunt approach that ignores the fact that some of the maven dependencies are, in fact, already OSGi bundles. The bundleall goal looks promising for this use case, but I have found that it creates many more bundles than I expect. Whereas my Embed-Dependency bundle had about 80 nested jars, the bundleall goal creates almost 200. I was hoping that it would create the same set of jars, but just "wrap" the ones that didn't already have OSGi metadata. This difference seems to be because Embed-Dependency uses rootProject.getArtifacts() while the bundleall goal recurses into each dependency. However, I don't fully understand the logic/difference here. Any insights? Or any other ideas for my particular use case? Thanks, Will -- View this message in context: http://www.nabble.com/-maven-bundle-plugin--differences-between-bundle-and-bundle-all-tp23511135p23511135.html Sent from the Apache Felix - Users mailing list archive at Nabble.com. --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
