Is this inspired by what Chris Okasaki calls persistent data structures in his marvelous little book?
Once you have a handle on the data structure, you can perform any pure operation, which is intended for that data structure and your handle will still point to the same data structure, because "modified" versions are not really "modified" as in someone performing a mutation on the data structure, but actually new instances of the data structure, which might share parts with the original one.
This property is very nice to have for working with the data structure.
I bought the book at one point during my Haskell years but found it too dry for my taste at the time.