Currently, the UDF can only allow yielding outputs to one output port. A motivation example is a UDF producing a chart as well as a table, which needs to be sent to different downstream operators (i.e., send the chart to the visualization operator, send the table to other processing operators).
The engine supports multi-output ports, but we need an interface in UDF for users to write code to specify an output tuple to a specific port.
Currently, the UDF can only allow yielding outputs to one output port. A motivation example is a UDF producing a chart as well as a table, which needs to be sent to different downstream operators (i.e., send the chart to the visualization operator, send the table to other processing operators).
The engine supports multi-output ports, but we need an interface in UDF for users to write code to specify an output tuple to a specific port.