Do you have a plan to support one-to-many node?

[luoling1993]

Now, the langgraph support many-to-one node, a node allows multiple nodes to use it as an end node, however, multiple nodes are not allowed to use it as a start node.
This design cannot be implemented in the following scenarios:

1. Multiple operations are required after a node, which can be done in parallel to reduce time
2. A node has multiple judgment logics, and different combinations of judgment results have different subsequent logics
3. Similar to listener mode, which logic fires depends on something in the state

[hinthornw]

Hi @luoling1993 - one-to-many (fan-out) is a great idea! As far as I can tell, we already support it in LangGraph via conditional edges (see doc on branching)

"Listener mode" is done by defining the firing logic in the outbound conditional edge that would trigger any of the target node(s) in parallel.

Could you elaborate on what is still missing?