Dear @mlauber71 ,
I am not exactly sure what that does… Do I understand correctly that this work with a collection of KNIME tables on disk?
I would like to keep the data in the workflow itself. I can hack together components which kind of do this:
It starts with 2 tables:
and
By using a sequence of GroupBys and Column Aggregators it is possible to gather all the data into a single Collection cell:
These cells can then be concatenated:
…and subsequently unwrapped again into a full table:
But I think I can hear the Java memory management system / garbage collector scream in agony when I combine lists with different element types in this way 
Naively I don’t think that it should be hard to send whatever output the Table Writer node is creating to a table cell instead of a file, which would probably be more efficient than the method above.
Best
Aswin
KNIME_project14.knwf (61.3 KB)