Ability to read and write variable values to an in-memory data pipeline as part of aggregation pipeline stages. As the documents are streamed through the aggregation pipeline stages, this functionality can open up enormous computational capabilities, already loving the conditional constructs switch/case,cond and added regex support.

Scenario: A document can mark its index position before and after sort stage and calculate its offset.

Understand that streaming behavior may vary on sharded clusters, but then onus will be on end-user how he leverages the feature, mongodb need not promise on any sequencing behavior, just an ability to leverage this stream based sequencing by end-users where-all possible.