When the Slice Loop End node encounters a column that is not sliced (e.g. it was created within the loop) it fails with a NullpointerExeption, this is not informative for users and should be changed.
See the attached workflow: SliceLoopFail.zip
In general, the matching of columns back to the input is too primitive (if for example the order of columns changes within the loop, it will result in unexpected results).
When the Slice Loop End node encounters a column that is not sliced (e.g. it was created within the loop) it fails with a NullpointerExeption, this is not informative for users and should be changed. See the attached workflow: SliceLoopFail.zip
In general, the matching of columns back to the input is too primitive (if for example the order of columns changes within the loop, it will result in unexpected results).