I see a few options to do so:
- use a sql recipe like this: SELECT ... FROM foo JOIN bar ON 1 = 1
- create a column in each dataset (called "key" for example) and with a constant value (1 for example). A visual JOIN recipe, joining on this column "key", will do the trick.
- this is also doable in a Python or R recipe.
As always with cartesian product, estimate the size of the output dataset before building it.