Missing a test that verifies the AccountInfo's data length is correctly enforced to exist in RW data. The length is translated correctly but we have no way to ensure that it will continue to do so over time.
this is the natural behavior of node-hid but we should backport this behavior to all other transports.
basically it's a failsafe for race conditions.
if you open() a transport, you are responsible to close() it in a finally, if an open() happens during that time it should throw if it concerns the same device (e.g. by path id / by uuid, ... depending on the transport)