Rapport · Engineering case study
Reject the bad row. Keep the useful work.
A malformed record should not halt an entire import or cross a customer boundary. I built ingestion around isolated writes, explicit rejection reasons, and tenant-aware identity.
PythonImport pipeline
PostgreSQLRow-level savepoints
The decisions
Three decisions, the reasoning behind each, and what changed as a result.
Fault-isolated ingestionImport exampleOne bad row. Seven useful records survive.
The bad row keeps an explicit rejection reason. It does not need to erase unrelated progress.
Compare the boundaries to see the cost of coupling every row to one failure.
How it works
A quick guide to the interactive flow on this page.
Dynamic contentClick each step to see a different part of the story.
The outcome
33 passing integration tests.
The pipeline isolates bad rows, skips duplicates, and keeps tenant data separate. The recorded integration suite checks this behavior across 33 passing tests.