etl
etl copied to clipboard
Data diff: strict checks when no changes expected
What
When a PR body says "No data changes expected", make the data diff check strict (i.e. fail with a red cross on any differences).
Why / Why now?
- Today we ignore data diff because it is never green or red
- To make it useful, we should let it fail if no changes are expected
- This will help it catch cases where we didn't intend to cause changes