Any given workflow node will outfield a set of fields that can be used in the input of descendent nodes. Because these outputs are only generated at runtime, we use a templating system to keep track of what values a node will receive once these values are actually calculated. This document goes over how this process works in both the frontend and backend.


All block definitions will specify both the inputs it takes in and the output values it provides


The workflow hierarchy follows a graph-like structure. For a given node, we calculate what output fields are available to it by doing a BFS on its ascendant nodes. Once we have a list of all ascendant nodes, we can generate the available output field values along with its underlying key. When placed inside the input of another node, it will take its template form which has the following shape:



A workflow task in kafka will keep track of the mapping of template key to value pairs as they are generated. When a new node is being executed, its input fields will first have actual values replace the string templates before type validation is run.