fix(ui): preserve precise audit submissions and adopt delivered secrets
This commit is contained in:
@@ -131,3 +131,23 @@ Browser acceptance is summarized explicitly in evidence/sb10-12/acceptance.json:
|
||||
The first pass stopped at missing fixture-only AuthorizedManualFailoverService DI before Health could render. After adding the real authorization wrapper over a rejecting synthetic backend, the retained initial22-case JSON recorded19 passes, including requested dark cases that actually stayed light and a false coarse context; those do not establish dark/touch acceptance. Final broad progress20 contains actual light/dark measurements but stopped awaiting coarse initialization. Its six immediate End-key geometry failures were measured before Bootstrap's existing smooth document scroll settled (timing proof retained). Focused confirmation waits for the actual last treeitem and viewport geometry, passing6/6 without keyboard source changes; two additional contexts explicitly verify CDP native touch emulation (maxTouchPoints1, pointer:coarse) and pass2/2. Raw failures, progress, scripts and corrected measurements remain separate; no all-first-pass-success claim is made.
|
||||
|
||||
The disposable localhost5201 host renders actual components/real policies with synthetic roles and rejecting read-only adapters. It has no database, fleet/actor system, LDAP or COM connection and is not complete application authorization/route acceptance. G1 historical evidence is unchanged. SB00 remains pending the published shared Secrets.Ui1.0.1 contrast patch and complete consumer baseline; no local undefined-token alias or caption override was introduced. No deployment or production name inventory was performed.
|
||||
|
||||
## SB00/SB13–15: delivered integration and final precise-query presentation
|
||||
|
||||
SB10–12 reviewed source/evidence was committed as `9dbfe7bc`. This final batch adopts delivered Secrets.Ui1.0.1 +Abstractions0.6.6 with Theme1.0.1; the umbrella Secrets/Core and SQL Server replicator remain0.6.2. Initial restore hit the old HTTP feed listing; a task-local empty HTTP cache and no-http-cache restore succeeded. official-package-provenance.json records all three exact accepted package hashes and official Gitea source. delivered-runtime-graph.json independently matches all three DLLs in both UI tests and actual Host output against their nupkg DLLs, retains bUnit2.7.2/AngleSharp1.5.2, and finds no bUnit/AngleSharp/xUnit/NSubstitute/test SDK in the production Host graph. No local Secrets color alias was added.
|
||||
|
||||
**SB13 ownership:** theme-ownership.json records the split. Shared Theme1.0.1 owns light tokens/fonts/component geometry, but has no shared dark palette or preference key. ScadaBridge therefore keeps its supported-dark token/Bootstrap adapter, domain/backdrop colors, `sb-theme` key, blocked-storage fallback and no-inline-Accent MainLayout contract. No preference migration or dark-palette deletion was justified. The existing Host pre-hydration script was copied byte-for-byte into the disposable fixture; eight desktop/mobile dark/light/absent/blocked-storage checks pass at first recorded frame and after hydration. This establishes the existing script's behavior in that document, not a first-paint claim about a deployed complete Host. Existing real light/dark Monaco, fields, dialogs, buttons and rail proofs remain retained in earlier batches.
|
||||
|
||||
**SB14 behavior:** the native audit form retains common Channel/Kind/Status/Site/time/error controls plus Apply/Clear; Node and rarer searches/exact identifiers move into native More filters. Both exact GUID fields reject nonempty malformed values before constructing a filter or invoking the callback, preserve input, and expose field-linked visible alert text outside the closed disclosure with recovery instructions. Whitespace remains an intentional unconstrained bound. AuditQueryModel's lax API and repository semantics remain unchanged. The summary says Last submitted filters, includes only actual emitted constraints, remains visible when More filters closes and survives Clear while Unsubmitted changes identifies the new draft. It never claims the subsequent grid query succeeded. Instance/Script remain the existing UI-only inputs dropped by ToFilter, so the disclosure count says fields set and does not assert those inputs constrain a query. Existing MultiSelect SelectionChanged callbacks refresh parent draft/count state; the first bubbling-only premise failed its targeted regression and was corrected without a shared API change.
|
||||
|
||||
The two former tests expecting malformed IDs to be silently discarded were replaced with whitespace-empty compatibility tests and independent malformed-ID no-callback/preserved-input/linked-error/correction regressions. Additional tests cover submitted Target/Actor visibility, pending vs submitted state, child Node selection, Clear of every advanced value/error, and all preexisting channel narrowing/UTC behavior. Red3→green23 is retained; final filter suite is25, including the final fields-set wording. Three process zones (UTC/Indianapolis/Tokyo) each pass25. No assertions or skips were removed to force a pass.
|
||||
|
||||
**SB15 presentation:** existing semantic text inks/state labels/mapping remain unchanged from SB07. The failed trace marker and audit JSON rule are now thin1px treatments; the debugger budget notice retains its explicit message, 1px semantic boundary and readable fill instead of an extra4px side bar. Predicted/Live/Replay/Running/Pending/Parked/Completed/Timed out/Failed behavior and error-log-on-completed-step semantics remain covered by the existing flow tests. No domain abstraction, new palette literal or state mapping was introduced.
|
||||
|
||||
Final checks: full delivered UI suite1596/1596, zero skipped/failed (sb-final.trx); related filters/flow101/101; final copy-only filter25/25. Actual Host Release builds with0warnings/0errors using the same official package cache. The first isolated Host build lacked assets at its new artifacts path; that raw failure remains, followed by an artifacts-path-matching restore and clean build. No Host process connected to external services.
|
||||
|
||||
Final actual-component browser acceptance is explicit in sb13-15/acceptance.json:20 logical cases across precise audit filters, delivered Secrets metadata in the actual Scada light/dark shell, real confirm modal keyboard boundaries and full synthetic flow outcomes at1440/390/real200% where relevant; plus eight preference cases. The initial broad run stopped at a strict selector matching both failed/timed-out trace rows. The corrected run retained14/20 immediate passes: all six Clear checks read the UI before the server-render response. Focused confirmation waits for actual cleared inputs and pending state, passing6/6 with no production fix. Both raw matrices and the correction remain separate; there is no all-first-pass-success claim. All accepted widths fit, real zoom records720 CSS pixels/DPR2, delivered caption contrast exceeds4.5 in both schemes, manager fixture metadata contains no reveal control, technical disclosures/filter/Clear work, native modal Tab stays contained and Escape restores its trigger, pending labels remain opaque. Mobile dark audit/Secrets and touch topology screenshots were visually inspected; some raw screenshots intentionally show the pre-Clear response or a horizontally scrolled local metadata table.
|
||||
|
||||
The full source route baseline is44 page components/57 declarations (including aliases), plus the delivered /admin/secrets package route; final-route-source-baseline.json names every source and policy hash. Runtime coverage is representative actual components with synthetic services, plus full rendered/unit coverage—not a claim that every authenticated route was browsed in a complete fleet-backed application. Production name/tombstone inventory (legacy names over450), legacy DIM exact-target restrictions, deployment, LDAP/COM/fleet/data-store acceptance remain explicitly outside this local QA. No production name scan, value reveal, database mutation or deployment occurred.
|
||||
|
||||
Peer review found and reproduced an out-of-order callback case: Apply A could await a slow caller, B could finish first, then A could overwrite the Last submitted summary. red-out-of-order-submission.log preserves the deterministic failure. Summary/signature are now captured immediately before invoking the callback, preserving submission order without equating completion with success. The slow-A/fast-B regression and all26 filter tests pass; final reviewed full suite passes1597/1597. This supersedes the earlier1596 checkpoint; no additional browser round was needed for the caller-order fix.
|
||||
|
||||
Reference in New Issue
Block a user