Hi Toni, thanks for the information. We will drive this effort further and then decide whether we go for this approach. Depending on the outcome we might donate the code.
Could you give us an idea of why the introduction of the TestAddress was a major impediment for the Remote Container feature? In my understanding the TestAddress is basically a key that allows to address a test method declared in a particular test class inside a probe bundle, running in a particular OSGi instance. TestAddress can spawn multiple child TestAddresses if the same test is to be run on multiple instances. The TestAddress is also crafted into the Probe manifest so that it can be resolved in the remote instance. Conceptually TestAddress should be a combination of container/test class/test method? Did you consider using OSGi remote services for communication with the remote instance? I would give that a try to see how it feels compared to directly operating RMI. Thanks and regards, Markus _______________________________________________ general mailing list [email protected] http://lists.ops4j.org/mailman/listinfo/general
