Unexpected Behavior in updateState During 'Wait for User Input' Scenario

[hide212131]

I have created a Java Notebook based on LangGraph's 'Wait for user input - Simple Usage' page. However, after interrupting once and then receiving user feedback, functions like updateState and getNext do not seem to behave as expected.

I have uploaded the Java Notebook here. The expected value for graph.getState(config).getNext() is 'step_3', but it remains 'human_feedback' regardless of the config value.

[bsorrentino]

@hide212131 thanks for this valuable feedback.

You must change:

var updateConfig = graph.updateState(invokeConfig, mapOf("user_feedback", userInput), "human_feedback");

With

var updateConfig = graph.updateState(invokeConfig, mapOf("user_feedback", userInput), null);

the last argument to graph.updateState is asNode which applies the update as if it were performed by the specified node