Description of problem: Throughout our codebase, we have this basic assumption - if a configuration property is required, we assume it exists in the supplied configuration object. Now from a usability point of view, there should be no need to supply those unless their values differ from the default value specified in the configuration definition. This is of course not a problem when accessing RHQ through GUI, because GUI takes care of that, but there is a number of remote API methods that either directly or indirectly use a Configuration class in their parameters. A remote user shouldn't have to be required to slavishly repeat the default values on required properties. The "impact surface" of this bug is quite large - basically the whole remote API needs to be checked if the arguments of all methods don't specify Configuration (or objects having configuration as one of their fields, recursively). Version-Release number of selected component (if applicable): How reproducible: always Steps to Reproduce: See bug https://bugzilla.redhat.com/show_bug.cgi?id=734866 Actual results: NPEs and other unexpected effects Expected results: our impls should be clever enough to replace the missing properties with their default values Additional info:
Just a note that this is sort of an umbrella bug for the individual issues yet to be discovered in the individual remote APIs. The bug 734866 mentioned in the repro steps is merely the first BZ raised that falls under this umbrella.