Closed cpita-felix closed 5 months ago
going to close this PR. My intention was to rewrite dataset.sql
as a series of direct left joins between activities, but I realized that that join could explode and generate spurious tuples, which would make it much slower to execute, and would require significant special case handling for the aggregate
style relationships.
Refactoring
macros/dataset.sql
to reduce the number of references to the original activity stream CTE/ref passed into it. Should therefore reduce average runtime of the macro as CTE logic/disk reads are performed less frequently.