Custom nodes are pipeline nodes that are not processes. They are supposed to be lightweight and can be used, for instance, to convert or transform inputs/outputs of a pipeline. For instance joining the result of many parallel execution in a single list parameter. The current custom node code cannot work in v3 due to to the intrinsic distributed execution. To resurrect them, we need to discuss a bit : What are exactly custom nodes for ? How to support their features in v3 ?
Custom nodes are pipeline nodes that are not processes. They are supposed to be lightweight and can be used, for instance, to convert or transform inputs/outputs of a pipeline. For instance joining the result of many parallel execution in a single list parameter. The current custom node code cannot work in v3 due to to the intrinsic distributed execution. To resurrect them, we need to discuss a bit : What are exactly custom nodes for ? How to support their features in v3 ?