Synopses, compressed data structures that summarize the full dataset, are used in data management systems to reduce latency while minimizing disk access. However, most existing stream processing systems like Flink, Spark, and Storm do not support synopses (which are hard to adapt to distributed settings) as pipeline operators. To make it easier to integrate synopses into any dataflow system that supports window processing, Poepsel-Lemaitre et al. propose Condor. Condor is centered upon a model that generalizes synopses as stateful window aggregate functions and classifies synopses’ algebraic properties to enable parallel computation.