Repository navigation
Repair selected output dates from retained raw data - #548
Merged
harrisonoconnorhover merged 1 commit intoOct 1, 2026
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Replaying a pipeline reruns its full output publication. Add an explicit saved-graph repair operation for selected UTC dates, using the raw data already retained in BigQuery. The normal source extraction and saved progress remain untouched.
The same run lifecycle owns the repair window, exact graph identity, retry, cancellation, history, and replay. Single-task Cloud Run binds the selection to its native execution, rejects older runtimes before pipeline work, and conditionally restores Job configuration after completion. Each existing output replaces only its selected interval in a fenced transaction; crossing business keys fail before publication.
Validation includes actual-row transaction and rollback checks through a narrow SQLite translation, runtime no-extraction checks, HTTP authorization/revision checks, durable retry identity, and Cloud Run adoption/cleanup races. Canonical typing, Ruff, and generated Control contracts pass. The full local suite's only failures were 28 loopback socket denials; all 28 passed with socket permission.
This is an experimental source implementation. Live BigQuery/Cloud Run qualification, immutable paired artifacts, and the guided Druff interface remain outstanding. The operator guide and support status state the existing-table requirement, per-output atomicity, scan scope, and lack of historical snapshot semantics.