RDP_WORKFLOW Connections

These connections are used by transformers with type rdp_workflow and are instantiated with the SDL workflow runtime instead of a Kubernetes Job (like all other transformer types — see Instantiation). Unlike TRANSFORMER connections (which use environment variables for direct communication between containers), RDP_WORKFLOW connections pass typed data between workflow nodes.

workflow_accepted_types

Every RDP_WORKFLOW connection must define workflow_accepted_types, which specifies what data types the connection can accept. Valid types are string, number, double, boolean, map, and array. The map and array types can be nested (e.g. an array of maps).

"inputs": {
  "WORKFLOW_INPUT": {
    "conn_type": "RDP_WORKFLOW",
    "display_name": "Workflow Input",
    "workflow_accepted_types": [
      { "type": "string" },
      { "type": "array", "items": { "type": "map" } }
    ]
  }
}

Connection Validation

When two RDP_WORKFLOW transformers are connected, the engine validates that there is at least one overlapping type between the source output’s workflow_accepted_types and the target input’s workflow_accepted_types. If there is no overlap, the pipeline will fail validation.

RDP_WORKFLOW connections cannot be mixed with TRANSFORMER connections. Both ends of a transformer-to-transformer connection must be the same type.