If the agent gets into a particular state (i.e. bad module specified), the agent can be crashed by a simple MQTT command to update. This is due to several places in the code where callbacks are calling agent.setState(...) without a .catch or a try, await, catch.
These issues were uncovered while developing a separate feature but were super easy to get into but difficult to get out of.
Expected Behavior
Ideally, the agent should not be crash-able at all (should be recoverable by the interface no matter the state)
This obviously is a utopian goal but the items highlighted can easily be addressed.
Current Behavior
If the agent gets into a particular state (i.e. bad module specified), the agent can be crashed by a simple MQTT command to update. This is due to several places in the code where callbacks are calling
agent.setState(...)without a.catchor atry,await,catch.These issues were uncovered while developing a separate feature but were super easy to get into but difficult to get out of.
Expected Behavior
Ideally, the agent should not be crash-able at all (should be recoverable by the interface no matter the state)
This obviously is a utopian goal but the items highlighted can easily be addressed.
Steps To Reproduce
No response
Environment