That's the question; I understand your approach to be reasonable if there is actually something to do during start-up (such as guessing the dialect, initializing a schema etc.) but if this is not the case, I don't see why deployment should fail in this case. We also don't stop the application if the connection goes away after deployment after all.
But if always validating the connection during start-up also if no actual "task" requires it is an established pattern in ORM, then we probably should stick to it. Although to me, this creates a more tight coupling between application and store than required. But if so, it should be done consistently for all grid dialects then, e.g. via a method such as DatastoreProvider#validateConnection() which is called during start-up. Today it's to left to the dialect's choice whether this check is done or not.
|