[insertish]

• main
  • fix(yucca sdk): restore schema on import #1
    • fix(yucca sdk): keep retention policy when changing other settings #2
      • fix(yucca sdk): skip snapshot prune if warnings are present #3 👈
        • feat(yucca sdk): job queue #4
          • refactor(yucca sdk): store log path relative to state directory #5
            • feat(yucca sdk): mark interrupted jobs as incomplete #6
              • refactor(yucca sdk): drop run history and tasks #7
                • feat(yucca sdk): job status UI #9
                  • test(e2e): generate tests for job API #68

[futo-kritika]

Confidence Score: 4/5

[Medium risk] When the unreadable-source error is caught, summary stays undefined. The lines after the try/catch, including the "Finished backup to primary backend" log and anything later that uses the summary, may dereference it, and the review does not say it checked this. The diff also adds no import of ResticBackupCommandCouldNotReadSourceDataError to the test file. The risk rating rests on this being ordinary backup and retention logic that a revert undoes, although skipping the prune could let …

Approved

Summary

Skips backup retention pruning when warnings are present.

RepositoryService.createBackup now collects unreadable-source errors alongside integration-hook warnings and runs automatic forget/prune only when no warnings remain. Warning-bearing backups still finish through TaskWarningsError, preserving the warning task status. The integration test exercises this path with a mocked unreadable-source error.

What's good

• Keeps unexpected backup errors fatal while collecting unreadable-source errors as warnings.
• Adds a regression test asserting warning completion without invoking retention deletion.

Review context · 0/6 commands

• Commands run: none

_{Reviews (2) · Last reviewed commit: "Merge branch 'fix/retention-policy' into..." · openai/gpt-6.1-sol · $0.12}

[futo-kritika]

kritika: confidence 4/5 with medium risk at c1fb9f3.

[futo-kritika]

kritika: confidence 4/5 with medium risk at f616b93.