[masarray]

Root cause and scope

Actual BCUGE screenshots from #507 portable test show CSWI1, CSWI10, CSWI2 despite the "natural sort" claim. The previous comparer was only installed by a search overlay matching six DataGrid columns, but the real native Explorer grid has seven including Alarm and owns a separate native search. This PR attaches the comparator at the real XAML DataGrid Loaded event (with source-change refresh), not an unused search module. WPF ICollectionView only, no data/FCDA order mutation.

GOOSE right Inspector and Messages previously bound terse names and hidden IED/Event columns. Changes:

• Time width 116, visible IED and GOOSE ID, Event, DataSet and Details. Source/destination MAC, relative time remain in immutable stream detail/raw evidence and are omitted from the compact event grid rather than squeezing critical identifiers.
• Application ID displayed as actual APDU number in decimal (0x002C→44, 0x0021→33); no hardcoded 44.
• VLAN ID and priority use typed parsed wire VLAN fields, not splitting an existing formatted label.
• Separate status number and sequence number; expanded "Time allowed to live", "Configuration revision", "Simulation test" and actual allData "Number of DataSet entries". Keep exact raw source in snapshot and preserve UI scroll.
• Additional constrained, source-bound adjacent quality pairing for the F650 screenshot: named stVal + next q both from active matching SCL stream, exact normalized IEC LN.DO display owner, valid decoded quality, non-conflicting full LD refs; no guessed Good or reordering of raw FCDA.
• Synthetic regression validates original 7-column Live Grid hookup (not just comparer unit), active GOOSE XAML, numeric APPID 44 vs 33, VLAN/status/sq/Test/wire entry count, F650-shaped q pairing.

One commit atop #507, engine pin unchanged. CI auto-trigger only once; keep R10 mainline guard intact and DRAFT until Windows portable smoke and BCUGE field test verify real result. No merge to main/release. Sources: user screenshots and live field diagnostic.

[masarray]

Exact-head qualification — BCUGE GOOSE operator UI (10 October 2026)

Latest HEAD: bccf18e33cf68289e26cdeddaae0daf203750f96 (3 scoped commits atop #507, 0 behind; 11 files). Source baseline ARSAS #507 57f92d98 + immutable engine #157 8b0c0f7e. No engine, static-reporting, NTP transport, or command code changed.

• Build ARSAS Windows x64 #38043934364 — SUCCESS: exact-head Release compile, entire application regression suite, portable single-file publish, EXE smoke test, artifact upload.
• Portable Windows x64 artifact #11667356157: candidate for operator's next BCUGE comparison.
• Production Promotion Guard #38043934376: SUCCESS on exact revision after correcting two UI regression assertions and malformed projection source in earlier provisional commits.
• Smart Discovery Field Capture #38043934309: SUCCESS (separate PR134 field packaging route). Golden Provenance, Golden Budget, Repeat-Run Stability, Merge Execution, SV evidence: SUCCESS.
• Mainline Readiness #38043934347 is still CLOSED/FAIL as designed pending unrelated R10 physical acceptance. Do NOT disable the gate.
• Previous commit attempts f2d0725 and 12df86f had failing regression/syntax and are not valid artifacts. Only use bccf18e or its run above. No manual workflow reruns or engine changes.

User field acceptance required

1. Live Values CSWI1/2/3/10 natural order on real seven-column grid, not the old six-column search-overlay heuristic.
2. GOOSE Time HH:mm:ss.fff untruncated. Visible IED + GOOSE ID; original MAC and relative-time evidence retained in historical event object and inspector, hidden from compact timeline.
3. Inspector Application ID is APDU decimal (0x002C→44; 0x0021→33), VLAN ID and priority from frame independently, Time allowed to live, Status number, Sequence number, Configuration revision, Simulation test and exact allData entry count.
4. GOOSE1/2 four actual wire members remain intact. Engineering stVal/q collapse only with exact/unique typed model evidence; no guessed Good. Test both actual frame/inspector and history.
5. Event cross-stream correlation remains conservative; independent or unmapped publishers do not disappear. Preserve every packet in raw capture and explicit retransmission mode.
6. BCUGE MMS Static DataSet stays report-only with previous 67/67 display, six exact RCB ingresses, and no cyclic polling.

PR stays DRAFT pending BCUGE screenshots confirming above. No main merge/release.

[masarray]

Disposition: Superseded — administrative consolidation of a verified commit-ancestry milestone, not a merge into main or physical acceptance.

Canonical replacement: #543 (head 6285867780b8ea79c231d72a3819e9e5f99748d2; stays OPEN for staged integration). This PR head bccf18e33cf68289e26cdeddaae0daf203750f96 is an exact Git ancestor of #543 head: GitHub compare old-head...canonical-head = ahead, ahead=227, behind=0.

Unique delta: IEC natural sorting and GOOSE engineering/event-inspector readability; 11 changed files, including GooseCanonicalEngineeringTests.cs, GooseNoviceInspectorRegressionTests.cs and NativeExplorerNaturalSortRegressionTests.cs. No original commits from this PR are omitted from #543's commit ancestry; the original branch and historical patch/tests remain recoverable. The final tree's semantic equivalence and physical qualification are not claimed by ancestry alone: these are follow-up requirements at the canonical tip.

Safety: Close only the redundant intermediate PR conversation, do not delete any branch. No code edits, no engine repin, no change to main, stable v1.6.42, checksums or safety/CI gates. This particular PR is not the remaining integration gate.

Follow-up: Retain typed GOOSE semantics, true wire provenance and regression tests; reconcile with any newer GOOSE authority before a physical/mainline promotion. Track unfinished acceptance and missing behaviors via #543 or the later authoritative integration/qualification workstream; do not merge #543 to main without required gates.