A customer would like the ability to retrieve only the changed portion (or at least have it indicated) for incoming mutations.
The use case is to track changes which happen on document much like change data capture (CDC) for databases where INSERTS, UPDATES and DELETES can be captured.
Today this is not possible with eventing as you only have access to the new (whole) mutation, there is no delta or previous version to compare to.
You could implement this by creating a second bucket and using eventing to store the old versions of mutations, then compare these in the function.
Unfortunately this means your storage requirements for your cluster immediately double and with de-duplication you may receive delta mutations which don't actually represent which individual mutations affected a document.