You can implement routing soon. The developers just presented this logic in a meeting (I’m not sure if it’s already live in production, but if not, it should be available within the next couple of days).
A node that ignores an error will always return a single value: NULL. All routing will be based on this, with two branches: node!=null and node=null.
This approach allows for proper handling of such cases. For instance, you can process an alternative branch, make a different request, or redirect the result to the current workflow