Files
lmxopcua/docs/ui-consistency/evidence/ot10-trees/README.md
T

26 lines
4.7 KiB
Markdown
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
# OT10 loaded-tree filtering and root recovery
Frozen source scope is GlobalRaw, GlobalUns, UnsTree plus TreeRecoveryTests (hashes in source-freeze.json). Both pages retain their exact ConfigEditor route policy. Create-a-cluster next step targets existing /clusters/new with the same policy. Raw retains prerender:false and auto-expands only eager Enterprise roots; it adds no child fetch. UNS filtering remains a case-insensitive substring over loaded names and ancestor context, with no new search endpoint or lazy tag fetch.
Root loading, failed read, confirmed no clusters and loaded-name no-match states are distinct. Retry is dispatcher-side single-flight. Failed retry returns focus to its control; Clear returns focus to the associated UNS input. The field describes its loaded-only scope; the helper sits outside the uppercase panel heading. Full escaped failure text uses existing native DiagnosticDetails. Page disposal cancels its view-owned initial/retry reads and ignores late results. Existing mutation/caller reload methods are unchanged here: their presentation-generation follow-up is a separate required OT13 batch.
The actual 195px browser case exposed an unbounded existing UNS action row (241px document). The loaded UNS region now owns horizontal overflow, has a contextual accessible name and keyboard focus; no virtualization or data loading was introduced. No global CSS change is included.
## Actual checks
- red.log: 4 behavioral failures before implementation (missing retry/no-match contract). first-green-attempt.log records 3 pass/1 failure from assuming bUnit wrapper disposal was synchronous; the deterministic dispatcher disposal correction is explained below. A separate concurrent Fleet/Hosts test compile failure briefly blocked an intermediate run and was corrected by its owner.
- tests.log/tests.trx: 13 passed, 0 failed/skipped (5 new recovery/filter cases plus 8 existing RawTreeRows tests). Includes repeated failed retry then success, no false empty state, pending read reentry guard, cancellation before releasing the delayed result, no lazy calls, loaded-only no-match and clear/unchanged read count.
- Command from OtOpcUa root: `python3 infra/tools/ui-consistency/with_private_sql.py --state <private-state-path> -- dotnet test tests/Server/ZB.MOM.WW.OtOpcUa.AdminUI.Tests/ZB.MOM.WW.OtOpcUa.AdminUI.Tests.csproj -c Release --no-restore --filter "FullyQualifiedName~TreeRecoveryTests|FullyQualifiedName~RawTreeRowsTests" --logger "trx;LogFileName=tree-recovery.trx" --results-directory /tmp/ot10-accepted-results`.
- fixture-build.log: actual component host builds with zero warnings/errors. Its proxy allows synthetic root reads and an empty equipment-picker context, refuses lazy/mutation calls and records refusals. No actual fleet, SQL or configuration write occurs in browser cases. All test discovery uses the explicit private loopback SQL runner.
- matrix.json/confirmation.log: 8/8 actual browser cases: Raw/UNS ×1440/390 and real200% (720/195 effective widths, DPR2). Checks failed-retry focus, escaped diagnostic, successful recovery, linked scope hint, no-match, Clear focus/value/loaded nodes, zero unexpected lazy/mutation calls and no page overflow/errors. Screenshots use viewport capture; tall viewport at200% avoids the earlier scrolling/capture framing issue.
## Honest failure record
The first harness attempt selected ancestor ThemeShell details as well as the target disclosure; harness-selector-failure.log preserves that strict-mode error. first-matrix.json/log preserve the first complete8case run: Raw4passed; UNS4 had case-sensitive CSSuppercase helper checks and immediate async-Clear assertions, plus the real195px overflow. The first confirmation then used global querySelector(input), which selected a hidden rail form input; global-input-harness-timeout.log and contextual-clear-probe.log show the selector mistake and actual correctly cleared/focused field. The final harness uses the control’s actual ID and waits on its value/focus; no Clear product behavior was changed to address that harness error. No failures are counted as passing.
The test fixture calls the component’s public Dispose on its dispatcher for deterministic pending-read cleanup, rather than assuming the bUnit wrapper’s synchronous Dispose has completed component disposal. This preserves the cancellation assertion. DispatchProxy fakes cover both root interfaces without changing the existing Raw fake’s throw-by-default contract.
## Remaining gates
Whole-route/full-suite/package/runtime and combined fields/dialogs acceptance remain OT14. GlobalUns modal replacement/reload generations remain OT13. No claim of generic tree search, complete lazy-tree inventory or production connectivity. Roll back this scoped source commit for a regression; there is no data migration or deployment.