Repository navigation
Conversation
|
No actionable comments were generated in the recent review. Walkthrough
Merge Risk:
|
…covery Retain pending checks with attempt and reset-generation fences, provide a storage-worker fallback for rejected rechecks, and retry counter cleanup without repeating rewards. Preserve votes arriving after the party boundary and exclude those votes from the preceding party's eligibility. Restore pre-dispatch state after commit or broadcast failure, retain failed restoration for subsequent recovery, enforce current enabled/period limits, and report force requests rejected while another party is pending. Includes the accumulated-vote regression from a2cd253 (PR #1755), together with recovery, ordering and scheduler regressions.
Purpose
Add a deterministic regression test for the pending VoteParty recheck identified during review of #1750.
The test starts a shared-MySQL VoteParty at a threshold of 3, leaves stored-recipient preparation pending, then records three additional votes and their normal party checks before allowing the first party to complete.
The required behavior is that the remaining three votes retain progress ownership and trigger a second party without requiring a seventh vote.
Scope
Test-only change:
votesAccumulatedDuringStoredPreparationTriggerNextPartyWithoutAnotherVoteExpected result
The original pending-recheck failure is already addressed in published #1750 head
06cb888b2be2c64cbacb1cc7e220b6bfd81d7bde. That commit contains this named regression with additional personal-eligibility assertions; its recorded focused validation passed. This PR remains a separate test-only change on its existing stacked branch; its older base snapshot does not establish a failure in current #1750. The required passing behavior remains:Base
This PR intentionally targets
fix/voteparty-shared-recipient-enumeration(PR #1750's branch), notmaster, because the regression exercises the new stored-preparation code introduced by #1750.Summary by CodeRabbit