Gents,
Over the last few months, I've been converting a large enterprise java application into a set of OSGi bundles. The goal is to have the bundles run in Karaf and perform the same things the previous application did (read files, communicate via JMS, etc). So far, I've successfully converted the application into a large set of bundles, and have successfully deployed them into Karaf 2.0.0 via a features.xml file. Each of the previous .jar files had the entry in the Manifest.mf file pointing to each .jar's main class in order to make them executable from a command line. Now that I've got these deployed into Karaf, of course those main classes aren't starting up the bundles. So, my question is, what is the most efficient way of starting these bundles so they will work properly in Karaf. One of the bundles reads files from the file system, and most of the communication between the major bundles is via JMS (a holdover from the java app). I have not created bundle activators, as I was told (likely incorrectly), that this was unnecessary if I use Spring DM, but that it required some sort of configuration file. After further reading, it appears what they were referring to is an OSGi 4.2 blueprint.xml document in each of the bundles. Is this correct? What is the best way to convert a bundle that was previously an executable .jar file with the following start-up method signature into a working bundle: public static void main(String[] args) Should I convert the former executable portion of the .jar into a bundle activator and use a blueprint.xml file, or is there a better way to accomplish this? v/r, Mike Van
