This file implements dbplyr backend support for Databricks SQL warehouses, enabling dplyr syntax to be translated to Databricks SQL.
Details
cumprod() returns cumulative products as doubles and propagates SQL NULL
values to subsequent rows in the group. Use dbplyr::window_order() with a
unique ordering (including a tie-breaker where needed) for reproducible results.
The translation collects and multiplies each cumulative prefix, so large
partitions can be expensive.