ADR 462: PR 88 Shell Return Correction

• Status: Recorded
• Operator approval: Not applicable: nonarchitectural task record under ADR 461
• Date: 2026-08-13
• Context:
  • Sonar reported that fixture_size ended with an implicit shell-function return around scripts/test-fixtures/lib.sh:12.
  • The fixture helper must retain the byte-count pipeline output and its exact exit status while satisfying the repository’s fail-closed Sonar posture.
• Decision:
  • Return the immediately preceding byte-count pipeline status explicitly with return "$?".
  • Extend the existing fixture-integrity shell test to prove the successful byte count and preservation of a nonzero pipeline status.
  • Treat this as a nonarchitectural corrective task record under the status semantics approved in ADR 461; no architectural decision is introduced.
• Consequences:
  • Sonar can verify an explicit function return without weakening, excluding, or suppressing analysis.
  • Callers continue receiving the same output and success or failure status as the pipeline produced before this correction.
• Follow-up:
  • Require the focused fixture shell suite and repository policy gates to remain green.
  • Confirm the refreshed PR 88 Sonar analysis no longer reports the explicit-return finding.

Task Record

• Motivation:
  • Repair the current PR 88 Sonar finding while preserving maximal scanner strictness and fixture-helper behavior.
• Design notes:
  • return "$?" is adjacent to the pipeline so no intervening command can replace its status.
  • The regression test overrides wc only inside a subshell and requires the sentinel status 37 to propagate unchanged.
  • No scanner setting, quality criterion, runtime architecture, or public interface changes.
• Test coverage summary:
  • Run just test-fixture-scripts for focused success, failure-status, integrity, download, and probe guardrails.
  • Run the relevant repository policy and instruction-drift recipes.
  • Run just clean-test-fixtures after validation and verify ignored media fixture directories are absent.
• Observability updates:
  • No runtime logging, tracing, metrics, health, or event surface changes.
  • The focused test emits a precise failure message if byte output or status propagation drifts.
• Status-doc validation:
  • Rechecked the ADR catalogue and documentation summary; this correction does not alter product capability or operator guidance.
• Risk & rollback plan:
  • The primary risk is accidental exit-status masking; the sentinel failure regression covers that behavior directly.
  • Rollback is removal of the explicit return and its focused assertions, though that would restore the Sonar finding.
• Dependency rationale:
  • No dependency was added or changed; the test uses Bash function scoping and existing repository tools.
• Stale-policy check:
  • Reviewed AGENTS.md and .github/instructions/devops.instructions.md for task-record, fixture, shell-safety, and Sonar requirements.
  • No policy drift or contradiction was found, so no instruction file change is needed.