https://github.com/apache/iceberg/pull/16176
This is NOT my PR, but I think it's very interesting. I've frequently wanted these types of queries in spark to run much faster, and many of my users do as well. It had some activity but has gone stale. What do you all think about it? PR Description: This PR enables aggregate pushdown for queries with GROUP BY on identity partition columns. Currently, Iceberg supports pushing down aggregates (COUNT, MIN, MAX) for queries without GROUP BY, computing results from file metadata instead of reading data files. However, when a query includes GROUP BY, the pushdown is disabled even when the GROUP BY columns are identity partition fields. -- Aaron Niskode-Dossett, Data Engineering -- Etsy
