Repository navigation
Consume smart interoperability engine and unify Discovery/Open-SCL runtime planning #446
Description
Activity
Engine implementation has started in ARIEC61850 draft PR #145 (issue #144). The first slice fixes the BRCB
ResvTmswire type at the engine semantic boundary rather than adding an ARSAS/vendor workaround.ARSAS consumer changes will follow the engine gate: consume the qualified engine, then implement the common canonical-model/static-first acquisition convergence required here.
Coordination update: engine #144/#145 now has a source-neutral configured-static activation entry point plus a pure
CanonicalIedModelstatic coverage resolver. This is intentionally being built in ARIEC first so ARSAS can later consume one canonical static-first plan instead of adding another app-side heuristic.I am not modifying active ARSAS PR #425 (SCL association consumer) or PR #445 (canonical package CI). The eventual #446 consumer change should layer on top of those accepted/parallel authorities rather than fork them.
Coordination checkpoint:
I re-audited current ARSAS parallel ownership before consumer work.
mainise4133cfc6291cc0f0b0202c550801edb0c47015f.- PR fix(scl): use engine-owned smart association resolution for Open SCL #425 owns SCL smart-association integration and touches
NativeIec61850Client.SclAssisted.cs, SCL preparation, engine lock and parity/interop tests. - PR ci(p3): seal canonical Windows package and reuse exact tested bits #445 owns CI-P3 canonical Windows package/installer exact-byte reuse.
- Neither lane is being modified from Consume smart interoperability engine and unify Discovery/Open-SCL runtime planning #446 work.
On the engine side, #145 base lane is now repaired and CI-green at
302f7882.../ CI #737. The next transactional readback-proof slice is isolated as stacked ARIEC PR #146 instead of being mixed into ARSAS.I am intentionally not repinning
engines/ARIEC61850.lock.jsonor creating a competing ARSAS consumer branch yet: doing so would overlap active PR #425's engine-lock ownership. Consumer convergence will start from the coordinated post-#145/#425 base, not by overwriting those parallel lanes.Smart hot-path consumer preparation is now staged on branch
fix/446-smart-report-hotpath, based directly on active PR #425 head (not on main).The staged consumer logic:
- builds/caches the engine-owned CanonicalIedModel from Live Discovery or source-correct Open-SCL ingress;
- resolves selected points to configured static coverage locally (zero MMS traffic);
- asks ARIEC to resolve covered DataSets to concrete live RCB targets;
- uses targeted availability only when dynamic residual search is unnecessary;
- otherwise retains broad availability so dynamic reporting safety is not accidentally hidden;
- explicitly enables ARIEC's reduced-evidence configured-static policy, which still requires populated live DataSet + RptEna=false + no positive busy evidence and does not relax dynamic mutation;
- diagnostics report targeted RCB count/read budget/DataSet cache hits.
I have intentionally not opened/CI-promoted this consumer branch yet: PR #425 has a strict engine-lock/interoperability guard that requires its physically accepted #143 pin. Repointing that lock in a sibling thread would violate the teamwork boundary. The combined engine candidate needed by this consumer is CodeVerified at ARIEC integration head
c0609d5.../ CI #747; consumer pinning should be stacked/coordinated only after the #425 ownership gate is resolved.- added a commit that references this issue
on Oct 7, 2026 Engine P6 consumer-contract milestone is now staged in ARIEC61850 PR #150.
Important integration rule for ARSAS: an inventory projected from Open SCL may be used for structural/canonical planning, but must not be labeled operational live RCB evidence. The new engine preflight requires explicit
LiveMmsObservedauthority before it will resolve/authorize concrete runtime RCB targets and perform targeted availability.This is designed so the eventual #446 consumer becomes thin:
selected signal intent -> engine canonical coverage/preflight -> engine exact-target availability -> ARSAS presentation/runtime binding.No ARSAS files or engine lock were modified here because PR #425 still owns that surface.
ARIEC P6 contract is now software-gated: combined exact head
51fa7092e83472a125aae9a29f033d31f59e6281passed .NET CI #758.The future ARSAS consumer should therefore delete/reduce local IEC orchestration rather than add another path:
- selected signal intent -> canonical selections;
- engine source-aware static acquisition preflight;
- engine exact-target availability;
- ARSAS only binds returned plan/result to runtime/UI.
Engine now also supplies canonical structural report-inventory projection with no fabricated RCB01/live-state attributes, which is the intended replacement for ARSAS-side trusted-SCL inventory fabrication.
Still intentionally no engine-lock repin or edit to PR #425-owned SCL-association files from this lane.
- added 8 commits that reference this issue
on Oct 7, 2026 54 remaining items
**Native WPF IED dialogs UX PR #486 final exact-HEAD
56de3e2e0e82f01f04c49ca270d37f752e99f86f: 9/9 GitHub Actions SUCCESS; 1404/1404 regression tests passed; native Windows portable EXE artifact #11587495503, field-capture Win-x64 #11587595323. Restyled IED Actions + Add IEC 61850 IED dialog-only with high-contrast light surface, natural height/scroll clipping protection, reduced noisy copy and preserved action/validation contracts. No engine/RCB/GI/report/session code edited. Visual user QA pending, draft/no merge/release. Owns only dialog view, small dialog copy C#, isolated dialog styles and two offline tests; preserve P7.6C #485 as stacked dependency.P7.6C Oct 9 real diagnostic verification — startup observability validated, SOE continuity not yet qualified
Operator supplied two separate ARSAS native startup diagnostic reports, same app assembly
1.6.40+0871caebe886327b4e442ee329a7d9fe95a9b895(GitHub test merge, parents397d7836e5b071d4f3c6b375f9c51879c58eeb48and8d764128ccf764b20d27d178dfec64ec71581b81, i.e. #485 built on #484), same engine lock352c81e6a798635c6addcee0683235ca87ad416d. This does not include later native-dialog UX #486 HEAD56de3e2, so it is not visual QA for #486.Open SCD AP J vs Discovery IP: both connected, monitoring, exact source-neutral fingerprint
cb563f46ea1a61634ab67c6af4802d353d8d8f9caf8e793b7a89580525ffd206with same three semantic basis rows, selected static values 13/13, exact RCB traffic 2/2 independently, zero uncovered and zero cyclic MMS polling. Indexed BRCB: Open SCDRpt_ind01, DiscoveryRpt_ind02(excluded from semantic parity by design).New P7.6C evidence captured successfully in BOTH app processes:
Report continuity : ANOMALY OBSERVED • frames=3, sequenced=2, processUpdates=60, streams=2, findings=1, BufOvfl=0, untracked=0. BRCBframes=2, SqNum=2, firstSqNum=0, lastSqNum=0, findings=1, EntryIDPresent=0, ConfRevChanges=0; early warningsequence discontinuity (duplicate/replay): previous=0, current=0. URCBframes=1, SqNum=0, missingSqNum=1, no findings. Identical pattern across separate access paths suggests repeatable startup metadata behavior but is not proof of missing/replayed events. SqNum/EntryID are IEC 61850 report optional fields; missing consumer metadata could be absent on wire under OptFlds or absent in engine projection — not distinguishable in this diagnostic. Do not suppress SqNum 0→0 warning on assumption of GI, or infer EntryID progress from presence=0.Qualification limits: These are two separate ~30-second application startup captures, not a before/after reconnect test within an association lifecycle. Device card shows
AWAITING PEER INGRESSper process; cross-process semantic parity established by human comparison only, no in-app paired proof. Lack of BufOvfl alone doesn't prove no loss. Saved user endpoint is also shown as an IP assigned to a local NIC; do not assert a separate physical relay without endpoint provenance.Smart next evidence before altering engine/diagnostics: capture explicit report OptFlds (SqNum, EntryID, BufOvfl, ConfRev), raw InformationReport reason/segmentation where available, and exact per-RCB RptEna/GI/reconnect lifecycle timestamps at ARIEC authoritative decode/RCB boundary. Do not duplicate MMS parsing in ARSAS consumer. Then run continuous monitoring with timestamped before/after reconnect report traces and optional controlled non-safety SOE stimulus; no breaker commands. Keep #485/#486 draft, avoid speculative changes to static reporting or engine lock. P7.6B static semantic/report parity remains good; P7.6C diagnostic completeness demonstrably working but event-history continuity remains unqualified.
P7.6D — engine-decoded OptFlds provenance workstream / ownership (2026-10-09)
Draft PR #487 on P7.6C #485 exact head
8d764128ccf764b20d27d178dfec64ec71581b81(standalone sibling to IED dialog UX #486, not a rewritten UI baseline). Exact HEAD3f2bc75d9efb3bbe84c0808650b17f64bbda4808. CI pending.Root cause confirmed from pinned ARIEC engine source:
MmsReportFrameMapper.DecodeHeaderalready extracts report wireOptFldsasMmsReportOptionalFields.RawHexplus typedHasSequenceNumber/HasEntryId/HasBufferOverflow/HasConfRevision. ARSASNativeIec61850ClientdroppedOptionalFieldswhen mapping engine report frame to consumerNativeReportFrameMetadata; therefore diagnostics for real pairGR_X_7SX85couldn't distinguish wire omitted vs requested but not decoder-projected vs no OptFlds evidence.Narrow read-only fix: preserve nullable decoded option bits and a maximum 8-byte hex mask; distinguish unknown/omitted/advertised-but-undecoded counters per exact RCB and write bounded evidence to Copy Diagnostic. Keep conservative SqNum 0→0 duplicate/replay warning and no-event-loss assertion. Test missing, explicitly omitted, unknown, contradictory SqNum/EntryID, empty EntryID ambiguities, hostile hex, engine-owned decoding boundary. 5 files only (adapter, immutable metadata, continuity inspector, diagnostics, new tests); no MMS/RCB writes/GI/control/polling/SCL/engine pin/runtime value path changes.
Parallel ownership: #487 owns ONLY these OptFlds provenance adapter/inspector/diagnostics lines, #486 native dialog UX owns separate XAML and dialog source; coordinate stack integration without dropping either. Physical reconnect/SOE event-history and in-app paired-ingress proof not achieved; do not merge/release prematurely. CI budget: one atomic 5-file tree push, one minimal wording contract correction, now wait on exact HEAD, not repeated speculative commits.
P7.6D PR #487 final exact HEAD
3f2bc75d9efb3bbe84c0808650b17f64bbda4808: 11/11 CI WORKFLOWS SUCCESS, 1,408/1,408 ARSAS regression tests PASS, IEC 61850 interoperability guard green, portable Windows x64 single EXE, field-capture Windows x64, and Windows x64 installer all available. Engine-owned InformationReport OptFlds evidence is forwarded through existing adapter into immutable per-RCB Copy Diagnostics, enabling wire omitted / advertised-but-undecoded / unknown classification for SqNum and EntryID, while preserving field BRCB SqNum=0→0 warning and report-only acquisition. No engine pin, network, GI, RCB write, cyclic polling or UI changes. UX #486 is a separate sibling workstream (not included in #487 artifact). Physical OptFlds/reconnect SOE testing still required; PR stays DRAFT, not merged/released. Avoid new CI unless actual field failure or approved integration.P7.6D field qualification — real disconnect/reconnect pair (2026-10-09)
Operator supplied two independent application diagnostic files, each showing initial connect → stop monitoring → disconnect → reconnect → Static DataSet reporting re-arm within the same app process. Assembly
1.6.40+b35dbc821faaab706536e0358ec962f1a6cf2e80is the exact GitHub test merge of #487 HEAD3f2bc75d9efb3bbe84c0808650b17f64bbda4808into #485 HEAD8d764128ccf764b20d27d178dfec64ec71581b81, engine pinned unchanged at352c81e6a798635c6addcee0683235ca87ad416d. Native dialog UX #486 not included.After reconnect, Open SCD AP J: same 13/13 displayed, exact 2/2 RCB routed, 0 uncovered, cyclic process polling=0, semantic fingerprint
cb563f46ea1a61634ab67c6af4802d353d8d8f9caf8e793b7a89580525ffd206. Association-local 7 decoded frames (6 BRCB, 1 URCB), 1 finding, OptFldsDecoded=7/7, Unknown=0. BRCB switched legitimate available instanceRpt_ind01→Rpt_ind04, mask7880with SqNum advertised=6/6, decoded=6/6, EntryID omitted=6/6. URCBRpt_meas01, mask3880: SqNum omitted=1/1 and EntryID omitted=1/1. New-session BRCB warningprevious=4,current=0. Earlier first-session warnings5→0and0→0appear in historical logs but are not inherited into current snapshot, indicating reset isolation.After reconnect, Discovery IP: same semantic fingerprint, 13/13 displayed, 2/2 RCB routed, 0 uncovered, cyclic polling=0. Fast reconnect uses cached 532-signal model without repeating full discovery. Association-local 8 decoded frames (7 BRCB, 1 URCB), 1 finding, OptFldsDecoded=8/8, Unknown=0. BRCB instance
Rpt_ind01→Rpt_ind02, mask7880: SqNum advertised+decoded=7/7 and EntryID omitted=7/7. URCB mask3880: SqNum omitted=1/1, EntryID omitted=1/1. Current-session BRCB warningprevious=5,current=0(older first-session5→0and0→0in historical logs). Both BufOvfl=0, ConfRevChanges=0, no ambiguous/untracked metadata streams.Acceptance: reconnection, static coverage, report-route authority and ARIEC wire OptFlds provenance are now physically confirmed across both ingress paths. Missing URCB SqNum is IED OptFlds-omitted on the decoded InformationReport, not an ARSAS decoder failure. Empty BRCB EntryID is likewise OptFlds-omitted; there is no EntryID replay cursor in these reports. Remaining unqualified: what triggers within-new-session BRCB SqNum reset
4/5→0(e.g., GI/integrity/RCB activation), whether any SOE history was lost, long-soak behavior and in-app paired dual-ingress proof; don't discard warnings or infer event loss. Separate process logs both reportAWAITING PEER INGRESS; comparing hashes across processes is useful but not the same as in-app dual-ingress traffic proof. Discovery still reportsPrimary unresolved=2vs Open SCD 0 (control/inventory provenance issue), though runtime uncovered=0.Recommendation: next scoped phase is engine-authoritative, timestamp-correlated RptEna/GI/first-frame cause evidence and a controlled harmless source-event test with known event count if physical setup permits; never issue breaker controls. Avoid new coding/CI for wire OptFlds classification: the field evidence supports #487 implementation as designed. Keep PR draft; no merge/release pending event-history qualification. No customer raw files, IP inventory or sensitive diagnostic payload committed.
P7.6E workstream ownership + CI-thrifty evidence correlation
Draft PR #488, exact head
d1e6cb9351d8283e99cc8a898dd82531eda96fef, stacked on P7.6D #487 exact3f2bc75d9efb3bbe84c0808650b17f64bbda4808. No modifications to the independent UX #486.Field evidence P7.6D real connect/disconnect/reconnect: 13/13 Static DataSet rows, BRCB+URCB 2/2 routed per ingress, zero process polling, both OptFlds decoded. BRCB SqNum requested and decoded (mask 7880), URCB SqNum omitted (mask 3880), EntryID omitted on both; BRCB new-association warning 4→0/5→0 remains unqualified as GI/reset/replay/loss.
Engine-authoritative discovery (pinned ARIEC
352c81e6...): existingStartConfiguredStaticReportMonitorAsyncuses trueRptEnawrite, post-enable readback verifier, one-shotGI=truewrite; returned engineWriteStepsalready include attempted/success.MmsReportFrame.Values[].ReasonForInclusionis already decoded by engine. Consumer currently discarded write outcomes and frame GI/integrity reasons. #488 forwards existing data, never re-parses MMS.#488 maps exact engine static-activation outcomes into per-association per-plan evidence, then into existing per-RCB continuity snapshot; maps decoded GI/integrity reasons only when available, retaining UNKNOWN for missing. Warnings preserve 5→0/4→0, with truthful timestamp/correlation text not asserting harmless reset or no SOE loss. 6 new synthetic tests. Atomic one commit / one CI trigger batch, tests preflighted before branch publish. 7 files:
NativeReportMonitorModels.cs,NativeIec61850Client.StaticDataSetReporting.cs,NativeIec61850Client.cs,Iec61850MonitorRuntime.cs,Iec61850ReportContinuityInspector.cs,DiagnosticReportBuilder.cs,ReportActivationCausalityP76ETests.cs. Other threads should not overwrite these without reconciling SHA/PR, and must preserve static reporting authority and UI #486 separately.CI pending and physical GI/SOE qualification pending. PR draft, no merge/release, no control commands or report/RCB/GI writes added; engine pin unchanged.
P7.6E causal report evidence draft PR #488 exact HEAD
d9d11ea2a24f09f6a515c3b561daca2a373d69c1: 11/11 CI workflows SUCCESS, 1,414/1,414 ARSAS regression tests PASS (guard); field-capture Win-x64 artifact, native portable EXE, installer ready. Strict single-source engine evidence: existing configured-static ARIECRptEnaandGIWriteSteps (success/rejection/unknown), decoded per-report ReasonForInclusion GI/integrity, association-local per-plan lifetime and precise report received timestamps. No MMS parser, new RCB/GI writes, polling, controls, engine lock or values path changes. BRCB SqNum 4/5→0 warnings preserved; observed GI correlation cannot certify no SOE loss. Formatter selectively emits lifecycle detail when actual evidence exists, preserving bounded diagnostic regression. Exactly 2 commits/CI batches (first failed one valid bounded-size test, corrected production code), no reruns. Field GI/reconnect evidence still needed; draft/no merge/no release; UX PR #486 separate.GE Multilin F650 field milestone: canonical command object + evidence-graded rejection (2026-10-09)
User supplied F650
BCU_GE.iid, Open IID/SCL and Discovery IP diagnostic. One physical IEC 61850BCUGEF650/CSWI1.Posappears as three apparent commands in the native WPF Command Dock because ARSAS static control projection previously keyed companion reuse by (DataObject,DataSetReference,DisplayReference) instead of by canonical fullIED/LD/LN.DO. Report/DataSet memberships must remain separate, but Operate is a single canonical target.CSWI1…CSWI16are distinct objects; do not collapse to logical-node classCSWIalone.First Close on F650: wire-confirmed MMS
object-access-denied (3); operator confirmed BCU was physically still Local. The Local condition is field information not transmitted in generic MMS response. Later Close/Open on Open SCL, and Close/Open on Discovery, received MMS service acceptance + process feedback. Do not claim a specific Local/Remote/interlock cause on another relay based on generic access denied alone. Pinned ARIEC engine already decodesLastApplError AddCausereasons (interlocking, synchrocheck, blocked-by-mode, no-access-authority, other client, etc.) where sent.Draft PR #489 stacked on P7.6E #488. Current exact head
0a47560f478437e50b399f0deba7b33dc2cedbd7; first commit07c9e384f28b010f8d319eca158619ef97357ba2compiled with 1422/1423 tests passed, one legitimate stale source-text-only assertion in FAT test expectedControlSupportsOperatedirectly inside MonitorModels after fail-closed gate was centralized into Iec61850ControlIdentity. Follow-up commit updated this test to assert the actual central fail-closed gate remains intact; no test weakened. It also added prominent, explicit operator MessageBox on failed command only (no modal on success) showing engine IED-cause certainty/evidence + checks, preserving raw request/response in Diagnostics. HEAD CI pending.Patch covers upstream control-companion reuse by canonical IEC reference, legacy/cached Command Dock alias dedup and unique count, one ctlModel inspection per distinct object, per-IED/per-exact-DataObject atomic TryAdd/Remove single-flight guard to avoid duplicate Operate, shared operator result classifier for Quick Dock and detailed command window, classified diagnostics. Explicit AddCause is confirmed; bare MMS access denied is MmsServiceOnly and Local/Remote, authorization, interlock are suggested checks, not inferred. No engine/SCL/reporting/GI/poll/control wire changes; commands not actually sent by development work. Native-dialog UX #486 remains a separate parallel sibling and must not be overwritten in integration.
Needs exact-head regression/packaging, then GUI screenshot and harmless authorized bench test of negative Local and successful Remote operation. PR stays draft, no merge/public release without field qualification. CI budget target now two automatic batches total, no manual reruns.
GE F650 canonical control and failure-reason milestone draft #489 exact HEAD
0a47560f478437e50b399f0deba7b33dc2cedbd7: 12/12 GitHub Actions SUCCESS, 1,424/1,424 ARSAS tests, 292/292 FAT tests, native Windows x64 portable EXE, Win-x64 field capture. Real F650CSWI1.Posrepeated per DataSet now normalized at producer and Command Dock; IED + exact DO single-flight prevents concurrent duplicate Operate across aliases; ARIEC AddCause is surfaced as confirmed when supplied, generic MMSobject-access-deniedremains unknown-specific with Local/Remote only a troubleshooting candidate; failed commands show prominent operator modal and raw protocol in Diagnostics. No RCB/GI/reporting changes or live commands issued in development. Physical UI/authorized safe bench acceptance pending; draft/no merge/release. UX #486 sibling remains unmerged/untouched.P7.6F — GE F650 field-driven Command Dock dedup and command shout
User-confirmed first GE Multilin F650 BCU operation was denied because physical BCU was in Local, before switching to Remote. The historical wire response was
MMS object-access-denied (3)with no confirmedAddCause/ControlError; therefore a Studio client cannot truthfully assert Local from that MMS error alone, even when human field evidence establishes it afterward.Draft PR #490, stacked on accepted-code P7.6E #488 exact
d9d11ea2a24f09f6a515c3b561daca2a373d69c1. Current exact HEADc08273201542d4372d1f05bc8d4f8d6ef0685add. Scope limited toModels/MonitorModels.cs,MainWindow.xaml,MainWindow.xaml.cs, and newtests/ARSAS.Tests/CommandIdentityRejectionP76FTests.cs. Does NOT change ARIEC MMS/Operate/SBO/RCB/control dispatch, engine lock or Static DataSet reporting. UX dialogs #486 separate; coordinate integration instead of overwriting.Command identity: F650
BCUGEF650/CSWI1.Posappears across several DataSets (e.g. CSWI, MMXU_SOGI1, REPORT1), but full device-scoped object reference identifies one actual command. Command Dock canonicalizes/group-selects one stable representative per exact target while preserving all underlying DataSet descriptors and feedback signal rows; CSWI1 and CSWI10 remain distinct.User requested SHOUT: a polished, non-modal, top-right warning card for failed command with a single resettable 5-second WPF DispatcherTimer. Real IED AddCause mapped to human-readable titles:
blocked-by-interlocking→Command blocked by interlock, synchrocheck/authorization/operating-mode variants mapped likewise. On genericobject-access-denied (3)with no AddCause, card says IED refused and advises checking Local/Remote and access; it does not assert that Local or interlock was encoded by IED. Command details remain in Diagnostics. Multiple rapid failures replace same card; no toast stacking.CI: first branch HEAD
9831760failed WPF parsing due to corrupted model file patch (non-protocol scope). Restored the FULL originalMonitorModels.csfrom immutable base and re-applied only tiny LINQ dedup in corrective commitc082732; no source model sections lost. Merge Execution Guard: 1,424/1,424 passed, 0 failed, 0 skipped. Other exact-head CI/portable artifact checks still pending. Draft/no merge/no public release. Physical F650 with Local and Remote status, relevant interlock AddCause, and Windows screenshot of shout still requires qualification; no breaker operation requested.P7.6F Command Dock + SHOUT draft PR #490 exact HEAD
c08273201542d4372d1f05bc8d4f8d6ef0685add: 11/11 CI workflows SUCCESS, 1,424/1,424 tests PASS (run). Native portable Windows x64, field-capture Windows x64 available. One actionable row per canonical device-scoped CSWI.Pos across multiple DataSets; a single non-modal toast auto-dismisses after 5 sec, honors explicit IED AddCause ('blocked-by-interlocking'→'Command blocked by interlock', synchrocheck, authorization/mode), generic F650 MMS object-access-denied only suggests checking Local/Remote without inventing a proven cause. No ARIEC/control send/RCB/reporting changes, user screenshot and F650 physical QA pending, #486 separate, draft/no merge/release.G1/F650 parallel PR reconciliation — integrated draft PR #491 (2026-10-09)
Two existing parallel, separately CI-green sibling drafts must not be merged blindly: #489 canonical Command DataObject identity, reporting DataSet alias projection, ctlModel inspection and per-target single-flight MMS guard, plus engine-based AddCause/LastApplError failure classifier; #490 non-modal CommandShout overlay and alternative UI-only dedup. Both base on P7.6E #488.
Created new integration draft #491 at exact HEAD
3c6ccc84a9fe97ae6e87262cdfd969abefedf497, stacked on #4890a47560f478437e50b399f0deba7b33dc2cedbd7. #491 takes only the nonmodal UX concept/XAML from #490 while using #489's stronger model identity and existing evidence-graded classifier; removes blocking MessageBox in command failure path, adds one reusable dispatcher timer, safe bounded notice, and tests. Generic MMS access denied remains reason unspecified, not proven BCU Local/interlock/auth; operator confirmed hardware was Local at first F650 failure, but wire response itself does not identify that cause.Change discipline: precisely 5 files, 1 preassembled Git tree commit and 1 initial GitHub CI batch (11 workflows automatically started); no new MMS/control requests, engine pin, RCB/GI, polling, live reporting or physical command changes. #486 is separate native-dialog UX; #488 remains canonical report causality workstream. Avoid editing #489/#490 branches concurrently; use #491 for integration proof, keep draft until exact CI and operator GUI screenshot/bench acceptance. Earlier #489 and #490 independent successes do not validate their combined runtime; #491 CI is presently running.
P7.6F G1/F650 integration PR #491 (stacked on #489, compatible visual shout from #490), exact HEAD
3c6ccc84a9fe97ae6e87262cdfd969abefedf497: 11/11 CI SUCCESS, 1,435/1,435 regression passed, native portable Win-x64 EXE and Smart Discovery field-capture EXE verified. One commit/one CI batch. Keeps canonical physical DataObject identity, multi-DataSet membership reporting, per-target atomic control dispatch, and evidence-graded engine AddCause/LastApplError classification; replaces modal error MessageBox with one bounded, reusable-timer nonmodal CommandShout. No engine, command writes, RCB/GI, cyclic polling or SCL changes in integration. A generic F650 MMS object-access-denied does NOT prove BCU Local; show it as a user-supplied bench observation/troubleshooting check unless device provides explicit state/AddCause. F650 visual/operator bench acceptance pending; draft/no merge/no release. #489/#490 are overlapping sibling PRs; do not merge both independently and regress this reconciled integration.Current ARSAS handoff / verified GitHub snapshot — 2026-10-09 18:02 WIB
Status checked directly against GitHub; not an automatic promotion. Current
masarray/arsas main=7d8d8026ae96d45f44d47f361d115a437af88fbe(NOT equal to active F650 integration branch). Enginemasarray/ARIEC61850 main=e5deed1d8aa11d97991695c6e390baafea7ab797, but the ARSAS active interop stack is pinned to immutable engine SHA352c81e6a798635c6addcee0683235ca87ad416dinengines/ARIEC61850.lock.json(engine PR #153). Do not silently move this lock to engine main.AUTHORITATIVE NEXT SOURCE: draft integration PR #491 HEAD
3c6ccc84a9fe97ae6e87262cdfd969abefedf497, branchintegration/446-f650-canonical-nonmodal-shout, based on draft #489 HEAD0a47560f478437e50b399f0deba7b33dc2cedbd7. #489 is stacked above report causal evidence #488 HEADd9d11ea2a24f09f6a515c3b561daca2a373d69c1, which descends from the #480–#485, #487 interoperability/reporting stack. #491 purposefully reconciles alternative F650 sibling #490: it carries the visible nonmodal command shout on top of #489’s canonical identity, single-flight protection and evidence-graded failure classification. Never cherry-pick/merge all #490 changes blindly into #491: alternative UI-only dedup would compete with upstream semantic canonicalization.Exactly verified CI on #491 SHA: 11/11 workflows SUCCESS; Merge Execution Guard 1,435 regression tests passed, 0 failed/skipped. Native Win-x64 portable single-EXE #11611067624; Smart Discovery field-capture x64 #11610687668. Artifacts exist in GitHub (not a public release). Code/CI/packaging accepted; physical GE F650 GUI/command field qualification pending; mainline merge/release not approved.
F650 field basis (operator uploaded screenshot, diagnostic and
BCU_GE.iid): 16 distinct CSWI*.Pos controls. E.g. CSWI1.Pos appears in three different DataSets and CSWI10.Pos in two; old Command Dock rendered aliases as multiple rows. This is one exact IEC 61850 command target per full device-scoped DataObject reference, not separate Operate services because of multi-DataSet membership. Open IID/SCL and Discovery IP both received actual BRCB/URCB reporting without cyclic process MMS polling, but selected signal counts differ (53 vs 24): do not claim semantic parity for this F650 comparison. Discovery hit 6,400 MMS-name/64-page budget andPrimary unresolved=244vs Open IID20, a separate Discovery completeness issue, not proof of command dedup bug.Initial F650 CLOSE attempt returned MMS object-access-denied (3) while operator independently observed the BCU was Local; later commands worked after switching to Remote. MMS denial alone does not prove Local, interlocking, synchrocheck or auth. Explicit decoded LastApplError/AddCause can prove exact reasons. #489/#491 provide
Iec61850ControlFailureReasonconfidenceIEDExplicitAddCausevsMmsServiceOnlyvsConfirmedClient/other, visible bounded nonmodalIec61850ControlShout8s with single reusable DispatcherTimer, and detailed diagnostics. Do not add fake Local/Remote conclusions or automatic retries. #491 retainedIec61850ControlIdentitydedup andConcurrentDictionary.TryAdd/TryRemoveper-target atomic dispatch gating; no extra MMS Operate to aliases.Other workstreams not to lose: separate native IED dialog readability PR #486 HEAD
56de3e2e0e82f01f04c49ca270d37f752e99f86fis NOT included in #491 (independent sibling off #485); reconcile deliberately later, with exact-head integration CI and Windows screenshot. P7.6B static member 0/1-based fingerprint normalization and equal Open SCD vs Discovery field fingerprint on Siemens-family IED were physically tested; #484–#485 continuity inspector and #487 OptFlds provenance were field tested, #488 GI cause observation code CI-tested, but SOE/event-history continuity unproven. Do not convert completed source/CI evidence into physical acceptance.Next work order: (1) review active PR statuses/main/lock, read
AGENTS.md,docs/WORKSTREAM_COORDINATION.md, current issue #446 and #489/#490/#491 PR descriptions+diffs; (2) obtain user screenshot from the exact #491 build showing one CSWI1.Pos row and distinct CSWI2–16, correct alarm/notice, no UI freeze, test low/high DPI; (3) use only a safe/authorized commissioning IED for negative/positive control test, capture Copy Diagnostic, LastApplError/AddCause + actual BCU Local/Remote state; do NOT operate energized breakers for UI QA; (4) if true defect, patch minimal #491 stack only with regression and exact-head CI; (5) reconcile #486 UX separately without overwriting #491/engine; (6) only after all physical gates, intentionally consolidate stacked PRs and synchronize newer main, run one final comprehensive exact-HEAD CI. Do not merge drafts prematurely or run redundant CI.Engineering/CI contract: engine is sole MMS/ACSE/RCB/control decoder authority; Static DataSet report-only mode never enables cyclic process polling or dynamic writes; preserve ordered FCDA, source SHA/IED/AP provenance, multi-association isolation, DataSet/RCB membership, and control single-flight. Async WPF no synchronous network UI. Prefer one atomic branch push after source preflight + synthetic failure-path tests. Keep thread ownership explicit to avoid overwriting parallel work. Treat operator-provided
BCU_GE.iidand diagnostics as contextual field sources, not redistributable repo fixtures. This comment makes NO code or release status changes.2026-10-09 combined F650 + native-dialog acceptance candidate
Created draft PR #492, HEAD
e8b05349061b5bbe600e53bea5323926d0df365a, as one atomic commit directly on exact #491 HEAD3c6ccc84a9fe97ae6e87262cdfd969abefedf497.Reconciliation was evidence-based:
- UX: legible, compact IED Actions and Add IEC 61850 IED dialogs #486 and P7.6F integration: canonical F650 commands + nonmodal evidence-based reject shout #491 share P7.6C merge base P7.6C #446: auditable bounded per-RCB continuity evidence in Copy Diagnostic #485.
- UX: legible, compact IED Actions and Add IEC 61850 IED dialogs #486 owns exactly six native dialog/test paths.
- P7.6D #446: retain ARIEC-decoded report OptFlds provenance, without MMS reparse #487/P7.6E #446: correlate engine RptEna/GI activation with report inclusion reasons #488/G1/F650 #446: canonical IEC 61850 command targets + evidence-graded reject reasons #489/P7.6F integration: canonical F650 commands + nonmodal evidence-based reject shout #491 do not modify those six paths.
- P7.6F acceptance: reconcile F650 command stack with native IED dialog UX #492 therefore reuses the exact UX: legible, compact IED Actions and Add IEC 61850 IED dialogs #486 blobs instead of reimplementing UX or merging sibling histories wholesale.
- P7.6F #446: canonical CSWI commands and evidence-based rejection shout #490 remains a sibling alternative; do not merge its UI-only command dedup over P7.6F integration: canonical F650 commands + nonmodal evidence-based reject shout #491/P7.6F acceptance: reconcile F650 command stack with native IED dialog UX #492.
Preserved unchanged from #491: canonical full-DataObject control identity, DataSet membership provenance, per-target atomic
TryAdd/TryRemovesingle-flight, evidence-graded AddCause/MMS failure classification, 8 s nonmodal Command Shout, report-causality stack, Static DataSet report-only behavior, and engine lock352c81e6a798635c6addcee0683235ca87ad416d.Exact #492 diff: only
IpConnectWizardWindow.xaml,
SclSignalSelectionModeWindow.xaml,
SclSignalSelectionModeWindow.xaml.cs,
Styles/IedTaskDialogStyles.xaml,
tests/ARSAS.Tests/IedTaskDialogVisualRegressionTests.cs,
tests/ARSAS.Tests/WorkspaceModeSwitchTests.cs.At this checkpoint GitHub automatically triggered one CI batch on #492; Merge Execution Guard, native Build ARSAS, Field Capture and other smart-discovery guards are in progress. No manual rerun was triggered. Code integration is committed; CIQualified and PhysicalVerified are not yet claimed. Keep draft; no merge/release. GE F650 one-row-per-target + safe command rejection screenshot/diagnostic and high-DPI dialog review remain physical acceptance gates.
Parent engine work
Tracks ARIEC61850 issue #144:
masarray/ARIEC61850#144
Consumer invariant
ARSAS must treat IP Discovery and Open SCL as two bootstrap/evidence sources for the same IED.
Both paths must converge into:
ARSAS must not carry a second copy of IEC 61850 protocol semantics or vendor-specific workarounds.
Field regression motivating this issue
A user test with a 7SX85-class IED showed:
ResvTmswith the wrong MMS scalar type;The engine defect belongs in ARIEC61850; ARSAS must consume the corrected engine behavior and remove/avoid downstream divergent planning.
ARSAS implementation strategy
A1 — One model boundary
Introduce/strengthen one canonical-model access boundary for downstream runtime consumers. Discovery-built and SCL-built models may retain provenance but must expose the same semantic contract.
A2 — One acquisition intent
ARSAS supplies intent/policy only:
ARIEC61850 owns:
A3 — Static coverage first for every selection path
Individual selections must use configured static DataSet/RCB coverage before dynamic reporting or polling.
If one selected signal is contained in a configured static DataSet:
A4 — Common reporting path
Do not maintain separate "Discovery reporting" and "SCL reporting" engines. After canonical-model convergence both must execute through the same report planning and activation contracts.
Open SCL may use bounded online reconciliation rather than full rediscovery, but this must not create a separate runtime semantics stack.
A5 — Evidence-first diagnostics
Surface enough engine evidence to explain:
Acceptance criteria
For the same physical IED and equivalent selected signals:
must produce semantically equivalent:
For a test set of 6 BRCB-backed + 6 URCB-backed signals, both paths should converge to the same static acquisition classification when live state permits it.
No ARSAS vendor-name condition or duplicated
ResvTmstype workaround is acceptable.