# Final independent GB migration audit November 13, 2026. This final pass supersedes the earlier in-progress snapshot below. **No concrete remaining migration defect or GB evidence gap found.** The parent owns the actual gate transition. Independently checked `seven-project-verification.json` against raw before/after `UnitTestResult` multisets (including duplicate display names), current restored assets, current executable dependency manifests, and all seven output bUnit/AngleSharp/Css/Diffing hashes. Verified **5,879 passed, zero failed, 75 inherited external integration skips, 5,954 raw results**. OtOpcUa has 2,167 raw results; collapsing duplicate test display names into a dictionary would incorrectly count 2,104, so raw multisets must remain authoritative. HistorianGateway's TRX summary skip counter is likewise not authoritative: its 75 raw `NotExecuted` records have exactly the same identities as baseline and all remain integration/environment-gated. No Dashboard/bUnit test is newly skipped. - All seven actual test graphs resolve exact bUnit 2.7.2 / AngleSharp 1.5.2, with no old core/web package generation. Runtime dependency manifests and output file hashes match the stated package set. No tests were rerun by this audit. - Theme's previously noted clean-preparation evidence condition is resolved by `Theme/clean.log`, `clean-restore.log`, `clean-after.log` and clean-after TRX (70 passing). Secrets/Overview retain their clean-run proof and unchanged identities/outcomes. - ScadaBridge `f8c2851d` and OtOpcUa `a47ffb91` retain all assertion-bearing lines and identical raw outcomes (1,551 and 2,167 passing). Scope remains test dependencies/compatibility, with parser suppression changes backed by accepted patched runtime results. - HistorianGateway `1be3f5e` changes only seven Dashboard fixtures and their test dependency declarations. Assertions and real callbacks remain unchanged. The sole baseline failure becomes passing because separate commit `2bbe3a0` changes only four deployment-document/evidence files to short host names; its corporate-hostname guard is unchanged. Outcome: 1,434 passed plus the same 75 inherited integration skips. These skips remain runtime limitations, not claimed passes or migration exemptions for affected bUnit behavior. - BTDB `0f9980b` retains real policy registration and every assertion-bearing line, while adopting the correct renderer/double APIs and awaited async component disposal. Its 426 Manager tests retain identical identities/outcomes. The prerequisite SDK is task-local exact 10.0.105; no global.json change. Separate authorized `b12117a` pins Core Pkcs 9.0.21 to remove pre-existing audited vulnerable restore failure. Core 350 passing tests and then Manager's passing 426-test before-bUnit baseline are recorded; the original failed restore is not represented as passing. - Current restored production graphs for the four sister repositories contain no bUnit/AngleSharp/xUnit leakage. The BTDB Core/Manager Pkcs 9.0.13→9.0.21 security delta is explicit and confined to the prerequisite; no UI/CSS/API implementation has started. The root three production graph checks were independently verified in the prior pass. - No new blanket skip, weakened assertion, audit disablement or production UI source modification was found. The earlier pending-worktree and failed-baseline rows below are historical observations, superseded by the final evidence and explicit user workflow override. The applicable seven-project **bUnit migration** acceptance is supported. This does not certify the skipped live Historian/Windows/Galaxy functionality or any later Theme/UI behavior. Retain the explicit integration limitations in the parent's final gate record. --- ## Historical in-progress review snapshot # bUnit gate review and seven-project acceptance map Reviewed November 13, 2026 from `/Users/dohertj2/.codex/worktrees/scada-ui-consistency/scadaproj`. No tests were rerun; this pass inspected committed changes, raw TRX, restored/runtime graphs and copied assembly hashes. No production UI changes were made. Other four migrations are in progress; this is an evidence snapshot, not GB closure. ## Review of completed umbrella slices **No concrete source-code defect found** in Theme `9bda64e`, Secrets `f961042` or Overview `8232f3a`. - Theme changes only its test parser version/comment, from AngleSharp 1.5.0 to 1.5.2; bUnit remains exactly 2.7.2. No Theme component, CSS, public API, production version or behavioral assertion changes. - Secrets' eight test fixtures differ exactly by `TestContext` → `BunitContext`, `RenderComponent` → `Render`, `AddTestAuthorization` → `AddAuthorization`. A normalization comparison of each old/new fixture confirmed no other edits. Central bUnit/parser pins and a direct UI-test parser reference are the only dependency changes. No assertion, skip, role/policy, strict JSInterop or fixture fake behavior was removed or weakened. - Overview's two component fixtures differ exactly by context/render API migration, with the existing local InstanceCard Render helper explicitly calling `base.Render`. All existing assertions, fake time, pause/adoption/resume and disposal tests remain unchanged. xUnit generation, SDK and runner pins remain unchanged. - Raw before/after result lists were independently compared as multisets of `(testName,outcome)`, not only totals or the supplied summary JSON. Theme has 70 identical passing identities; Secrets 34; Overview 197. The separate Secrets/Overview clean-after TRXs match the same identities/outcomes. Zero failures or skips in these three slices. - Actual current `obj/project.assets.json` and executable output `.deps.json` resolve `bunit/2.7.2`, `AngleSharp/1.5.2`, Css beta.157 and Diffing 1.1.1. Current output AngleSharp.dll hashes match the exact resolved 1.5.2 net10.0 package file for all three. This corroborates executed render tests against the intended parser; a declaration alone would not. - Actual production Theme, Secrets.Ui and Overview restored graphs contain no bUnit/AngleSharp/xUnit package leakage. Each recorded transitive NuGet audit reports no vulnerable packages from its current sources. ## One remaining evidence condition for BU-10 Theme's committed `after.log` proves a restore/build/test and matching parser binary, but its evidence folder has no explicit clean-output / forced-restore log equivalent to Secrets/Overview's `clean.log`, `clean-restore.log`, `clean-after.log` and clean-after TRX. This is an evidence gap against BU-10's **clean restore** wording, not a discovered compatibility failure or reason to reject its passing 70-test runtime run. Record the actual clean preparation if already done, or include Theme in the final planned clean gate sweep. Do not label this condition satisfied from package declarations alone. ## Read-only seven-project map All paths in this table are relative to `components/ui-theme/validation/bunit-2.7.2/` in the umbrella execution worktree. Counts are raw TRX observations at this snapshot; later files must be reviewed before updating a gate result. | Project / task | Baseline evidence | After evidence | Current acceptance assessment | |---|---|---|---| | Theme / BU-03 | `Theme/before/before.trx`: 70 passed | `Theme/after/after.trx`: 70 passed; `resolved-graph.json`, `runtime-integrity.json`, `audit.txt`, `test-comparison.json` | Source/runtime slice verified at `9bda64e`; final clean preparation evidence condition above | | Secrets / BU-05 | `Secrets/before/before.trx`: 34 passed, actual before parser 1.5.0 | `Secrets/after/after.trx` and `clean-after/clean-after.trx`: 34 passed; before/after graph, audit, runtime dependencies/hashes and identity comparison | Slice verified at `f961042`; no new defect found | | Overview / BU-06 | `Overview/before/before.trx`: 197 passed | `Overview/after/after.trx` and `clean-after/clean-after.trx`: 197 passed; before/after graph, audit, runtime dependencies/hashes and identity comparison | Slice verified at `8232f3a`; no new defect found | | ScadaBridge / BU-04 | `ScadaBridge/before/before.trx`: 1,551 passed, zero failures/skips | No completed after TRX reviewed in this pass | Migration in progress; patched parser/suppression disposition, full after graph, production boundary, clean run and identity comparison pending | | OtOpcUa / BU-07 | `OtOpcUa/before/before.trx`: 2,167 passed, zero failures/skips | No completed after TRX reviewed in this pass | Preview replacement in progress; preserve xUnit v3 cancellation/Akka stack; full after graph, parser/suppression disposition, clean run and identity comparison pending | | HistorianGateway / BU-08 | `HistorianGateway/before.trx`: total 1,509, executed 1,434; 1,433 passed, 1 failed, 75 not executed | No completed after TRX reviewed in this pass | Baseline is not green. Failure is `Contracts.CorporateFqdnGuardTests.TrackedTree_CarriesNoCorporateFqdn` in pre-existing deploy docs/evidence. Preserve/report it and classify all not-executed cases; migration and complete acceptance remain pending | | BTDB / BU-09 | No completed before TRX reviewed in this pass | No completed after TRX reviewed in this pass | Baseline/migration in progress; real policy, navigation, row/cancel precision and full after evidence pending | The HistorianGateway baseline failure is not a bUnit upgrade regression: it occurs before the upgrade. It still prevents representing that whole baseline as passing. Any resolution must preserve the guard/assertion and be explicitly reflected in before/after comparison; do not bypass it to close GB. This pass does not decide the permitted scope of that independent documentation correction. ## BU-10 final assembly checklist The parent should use actual terminal evidence for each of these before marking GB complete: 1. Re-discover the seven authored bUnit consumers after migrations, including conditional/central references. Read actual restored assets and runtime dependency manifests: exactly stable 2.7.2, no old core/web generation or preview graph. Record the actual execution checkout/branch/commit; old WORKTREE-PENDING markers are historical after the user's workflow override. 2. For every project, compare raw before/after test identities/outcomes and runner totals. Explain pre-existing failures, parameterized count differences and each environment-dependent not-executed case. No unaccounted removed identities, broadened mocks, new blanket skip or weakened assertion is acceptable. 3. Confirm final clean restore/build/full-suite run and targeted behavior evidence for render/query/click/rerender, allowed/denied policy, strict JSInterop, asynchronous wait/disposal and cancellation as relevant. The small async disposal probe supplements real suites; it cannot substitute for a project's missing fixture. 4. Record current audit output without disabling NuGet audit or enlarging suppressions; only remove prior scoped parser suppressions after accepted patched parser and real render results support it. 5. Compare actual production/package dependency graphs to ensure test-only bUnit/parser/xUnit references have not leaked. Record exact parser/Css/Diffing/framework/runner graph and output artifact hash evidence. 6. Update `gate-status.json`, baseline/compatibility records and acceptance document from the completed seven results, removing stale worktree-block explanations where superseded. Leave GB open until all required acceptance conditions are met. Theme/API/CSS and app UI implementation remain prohibited before GB closes. No commits were created by this review pass.