And another interesting one: Is it possible that a

node

has a (dynamic)

Parameter

as output? E.g. I have multiple "normal" parameters defined which serve as input into a

process_params

node. That node should - dependend on the normal parameter inputs - output a single parameter which might serve as input to other nodes. Currently - by simply outputting that "parameter" value - this is by default a

MemoryDataSet

Why do you need it as a parameter? As you said it’s not a “normal” parameter and is an input to other nodes.

As it is handled as a

MemoryDataSet

I cannot say I need it. But in my use case - logically - that unit of information is a parameter and not part of any dataset. So it would be nice to treat it as a parameter type (pot. nice for ipython session/kedro-viz/mlflow extension,...)

It’s possible to do it via hook, tho we generally don’t recommend doing a lot of dynamic parameters, this makes pipeline hard to debug and reason. kedro-viz provide you a static view about the pipeline, when you add more dynamic behavior it starts to diverge from the view as well.

Thank you for your input. So you recommend to stick to a (Memory)DataSet for "dynamic information" which is not data itself, in general?

Yes, I guess the name of

MemoryDataSet

make it sounds more like a dataset, but it could really be just treated as a python variable. In general I also do this combined with some convention, i.e. for Dataset I use CAPITAL, and anything in memory just using a lower case, same as python variable name convention.