2× speed
Duckstring
DuckDB Data Engineering
What if pipelines ran on demand from the end, instead of on a schedule from the start?