Skip to content

Bind recurring billing periods to genuine Order, Payment and Shipment flows - #1182

Merged
rmcdaniel merged 4 commits into
mainfrom
feat/1069-order-periods
Nov 17, 2026
Merged

rmcdaniel merged 4 commits into
mainfrom
feat/1069-order-periods

Conversation

@rmcdaniel

@rmcdaniel rmcdaniel commented Nov 17, 2026 •

Copy link
Copy Markdown
Member

Recurring billing with a genuine Order now keeps each original period bound to the existing Order, Payment and Shipment routes. Payment-first and grace activation share the common financial states; grace access alone cannot reserve stock or start fulfillment.

The trusted Order-period grant is part of the subject authority. Original admission and its receipt are atomic; replacement bootstrap cannot omit or substitute the policy. Access requires the original timed capture and protected paid publication. A first charge after the cutoff is refused, delayed acknowledgement preserves an on-time commit, late money cannot reactivate access or fulfillment, and physical work retains its separate original budget. Future cancellation fences later periods while current cancellation/recovery selects the original Order and refund. The provider's timing capability and schema are checked before native admission or money dispatch; the existing BillingPeriod payment-descriptor API remains unchanged.

Validation on 504f3cbd:

  • Current hosted run: full unit 7,096 tests / 87,282 assertions, 49 declared skips; contracts, static analysis and all eight PostgreSQL shards pass. Exact PostgreSQL inventory is 3,455 cases / 34,081 assertions with 50 declared backend skips; all 17 new Order cases execute.
  • Current default-Maria Order class: 17 tests / 106 assertions pass. The completed immutable predecessor baseline plus this replacement class covers all 3,455 current feature identities exactly once, with 34,112 assertions and 39 declared skips. This is composite evidence, not a new full-current-head run. Database settings match and all captured resources are absent.
  • 185 focused unit controls, ECS/idempotence and full static analysis pass. Both timing-capability predecessor failures are retained as causal evidence.
  • Refreshed timing hints preserve all 220 feature files, original deadlines and coverage gates. The previous PostgreSQL timeout remains failed; the current slowest PostgreSQL job completes in 17m39s. The provisional new-class PostgreSQL estimate is documented separately from measurements.

The current PR's bounded MySQL smoke fallback does not include the changed Order class; complete actual-main MySQL/coverage qualification remains required before source delivery acceptance. This source/example increment addresses #1069 PAY-06/SUB-01/SUB-02. The full 33-scenario matrix, fresh-process/installed qualification, missing-root startup boundary and publication remain unfinished. Workflow #1130/2.29 stays held.

@sonarqubecloud

Copy link
Copy Markdown

@rmcdaniel
rmcdaniel marked this pull request as ready for review November 17, 2026 05:32
@rmcdaniel
rmcdaniel merged commit 36dcd4d into main Nov 17, 2026
32 checks passed
@rmcdaniel
rmcdaniel deleted the feat/1069-order-periods branch November 17, 2026 06:58
@rmcdaniel

Copy link
Copy Markdown
Member Author

This source increment is delivered on accepted main 9a6c8cb6cda887239ba65485d2b6e5a12f432151, after the metadata-only scheduling correction #1183. Its product and test bytes are unchanged from reviewed 504f3cbd; the final merged tree equals the tested scheduling successor.

Actual-main build 44031867483 passes:

  • 7,096 unit tests / 87,475 assertions.
  • All 16 MySQL and 8 PostgreSQL shards, each backend covering all 3,455 identities exactly once. The 148 provider/financial examples per backend include all 17 genuine Order-period cases without skips.
  • All 17 unit/MySQL coverage reports, 94.10% line coverage, the unchanged ratchet and external Codecov checks. All four actual-main workflows pass.

The original main build 44024863937 remains FAILURE: its combined MySQL shard reached the original 1,200-second limit. The correction isolates the new class using a conservative scheduling hint; the actual covered class then completes 17 tests / 106 assertions in 656 seconds. The longest final MySQL shard is 1,057.82 seconds and PostgreSQL shard 938.44 seconds. No test, coverage driver or deadline was weakened.

Default-Maria evidence is explicitly composite: the complete frozen baseline plus the current 17-case Order class covers all 3,455 identities / 34,112 assertions / 39 declared skips, with unchanged shared source and settings. It is not a fresh full-current-head or installed run. All captured local resources remain absent. Both owned topic branches are independently absent remotely and locally; the worktrees are clean and detached.

The delivered example binds each period to its genuine original common Order, Payment and physical Shipment under payment-first or grace, with original money timestamps, late acknowledgement, expiry, access, cancellation/future-period and refund fences. Full #1069's 33 customer requirements, remaining physical/partial-stock/fault scenarios, the missing-root startup boundary, installed upgrade/recovery/rollback and publication remain open. No release, #1130 or 2.29 action is selected.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant