Pax Runner: Any experiences with the in-process executor?

I am currently deciding which pax executor to use when running OSGi applications from the IDE (see http://paxrunner.ops4j.org/space/Executors for a list of available one's). There basically two choices for me when I want to run a process from the IDE with pax runner:

In-Process-Executor (Runs the OSGi process in the same Java Process as Pax Runner itself):

• PRO: Easy to attach a debugger to.
• PRO: Easy to kill the OSGi process, as开发者_StackOverflow社区 no second Java VM is started.

• PRO: Faster to start.

• CON: heavily limits paxrunner's capabilities to non-javaVM settings (that one is from the documentation of Pax Runner).

Default Executor: (Runs the OSGi process in a new Java process)

• PRO: enables Pax Runner to set various JavaVM settings.

• CON: Harder to attach a debugger to (needs some remote debugging setup).

• CON: Almost impossible to kill the OSGi process if Pax Runner doesn't do it properly.
• CON: Longer startup times as two JVMs are started.

So my question basically is, if someone experienced a scenario where Pax Runner's JVM settings capabilities were crucial and the OSGi process would not work when using the in-process executor. I have not yet found such an example, however I need to decide if I support in-process, default executor or both, so some real-world use case that makes use of JVM-setting capabilities of the default executor would really help me in making that decision.

If the reason you start Pax runner from IDE is for Testing, have a look at Pax Exam, which also (optionally) uses Pax Runner underneath. Then you don't need to worry too much.