[masarray]

New physical FAT #662 follow-up

User measured SIEBCU v1.6.40 PR662: directory=15,590ms, types=2,837ms, projection=2,327ms, total discovery=21,397ms, then priority initial MMS read 32/32 complete. Monitoring 18/18 reported from BRCB, 69 frames no sequence alert, 5 control flights 0.77–6.73s, all positive CommandTermination. Two-thirds or more of slow flight is native service time; generic UI animation is not the cause.

P0 fixes

1. ExplorerSignalName now invokes shared exact IEC phase semantic presentation only for true .phsA/.phsBC tokens, rendering A PhsA / PPV PhsBC while retaining nonphase CSWI1/XCBR1 positional labels and untouched MMS reference.

2. Facade command admission rejectWhenBusy: true: zero-wait on both per-IED OperationGate and StopGate, reject with explicit NOT SENT when competing lifecycle/Read All/inspection/Stop owns the lane. No FIFO command backlog, no implicit cancel of in-flight MMS, no out-of-order SBOw/Operate. Native exact single-flight, ctlModel/interlock/synchro, quarantine on ambiguous outcome, positive termination and report feedback unchanged.

3. Root discovery performance is separately addressed in engine ARIEC #186: peer-negotiated window bounded read-only RCB enrichment, preserving exact Kpi/protocol evidence. This consumer branch currently retains the previous pinned engine SHA until CI qualifies and deliberate integration; do not claim directory speed improved from two UI edits.

Qualification

Test verified display names in actual Device Explorer point object, nonphase no regression, command fail-fast contract and native safety assertions. Run exact SHA Windows CI. No merge into main; real SIEBCU command and discovery repeat run needed.

[masarray]

Physical FAT SIEBCU follow-up — measured vs inferred (2026-10-21)

App PR #662 physical EXE ddede34... (consumer SHA from #662), engine 9c7cb403... on 192.16.1.192:102. Scan logged LD=5, DO=654, DA=4629, RCB=66, datasets=2, raw MMS=6281, 191 discovery requests, negotiatedCalling=5. Timing: directory=15590.3ms, GVA types=2837.3ms, model=592.1ms, projection=2327.0ms, total=21396.9ms. Initial read after discovery: 32/32 succeeded, zero failures. Monitoring 18/18 on static BRCB, 69/69 sequenced report frames, zero sequence findings (NOT SOE completeness proof).

Five consecutive CSWI1 position operations from native report: click-to-result ~6751,4700,772,5506,4188ms, with engine total ~6724,4695,767,5502,4184ms. Thus WPF/UI overhead is only milliseconds for those samples, and a UI queue is not physically proven as the 4-6s bottleneck. First slow SBOw service=1611ms and Operate=4523ms; fastest SBOw service=27ms and Operate=304ms; all positive termination and confirmed status. The 400ms SBO late-application-error safety grace is retained. No artificial timeout cut, IED-side latency allegation or bypass of interlock/synchro checks.

#663 consumer changes (1) exact .phsA/.phsBC labels in ExplorerSignalName, preserving existing LN-instance display for already-contextual engineering labels; (2) fail-fast zero-wait command admission for busy per-IED lifecycle/Read/inspection/Stop, marked NOT SENT, versus silent FIFO; no extra Operate. This does not claim to fix native MMS service response latency.

ARIEC #186 exact engine d095b3ac... separate RCB read-only bounded concurrency by negotiated window (5 on SIEBCU). Consumer trial lock and ancestry guard are code-only; physical SIEBCU repeat capture must compare directory timing and exact KPI. Trial SHA requires Windows tests and PE-verified executable before use. Do not promote to main / public without field qualification.

[masarray]

Exact artifact and CI checkpoint — PR #663

Source HEAD 03c668cb2474943988c16be6fcee8c31e0b234f4, synthetic PR merge 31de2639a10e1f91f4f217ef75e730f1bfc4acbd verified zero file differences from source HEAD. Engine locked to d095b3ac9bdf6687f7ab9407f7e28eca9c168c94 from ARIEC #186 (5 commits ahead of prior 9c7cb403..., exact diff restricted to smart discovery scheduler, policy and tests). Native ARDIREC remains locked at 7dbc7149db668126a493ce338264363a6ec5ec00.

Build run #39556650387 success; 2508/2508 ARSAS tests PASS, 24/24 pinned engine project tests PASS, 6/6 packaged regression PASS, native package smoke PASS. PE x64 direct EXE ARSAS-1.6.40-win-x64-portable.exe SHA-256 ae72327c812a8e2cf320f5c332c806ec45fee557f27d2044c2cb68f33f12881b, 79,127,539 bytes, compared directly with embedded build-identity receipt. Artifact ZIP #12326685184 GitHub digest d38f920c7da6400956e15b9d9b34dea95a6171e67c17a2205ff47593a82cac8b. Operator received extracted EXE directly via chat, NOT renamed ZIP.

CI transparency: 14/15 workflows success. Only CI P0 Workflow Integrity fails due to pre-existing repository topology > old documented budget: actual workflows=32 vs limit 31; actions checkout=39 vs limit 37. These ceilings were deliberately NOT raised, workflow deleted or bypassed. This blocks full 15/15 green release claim, but it is not a failing Windows compilation/test or IEC protocol regression. No main merge, no production promotion.

Physical FAT still required: record exact discovery directory/types/projection before/after on same 192.16.1.192 association: prior total 21.397s, directory 15.590s and negotiatedCalling 5. Verify all 66 RCBs, 2 DataSets and 4629 attributes preserved, zero duplicate request or accidental RCB writes. Confirm A PhsA / PPV PhsBC and position LN instance names. Command CSWI1 must retain explicit user confirmation, SBOw→Operate→positive CommandTermination and report-based feedback. New app non-queue gate rejects competing lifecycle/read/inspection as NOT SENT; diagnostic shows actual WriteControlAsync service latency varied from 27ms/304ms fastest to >2sec stages, so physical command service time improvement is NOT asserted. Physical evidence must distinguish MMS service wait from UI queue before a speed claim.

[masarray]

D28 maintainer-authorized archival: older trial PR superseded by the latest accepted working evidence, not merged. The operator will not redo field captures to qualify this historical branch. It is therefore withdrawn as an integration candidate, not treated as a new unresolved capture obligation. Historical HEAD 03c668cb2474943988c16be6fcee8c31e0b234f4 and full diff remain accessible; any unique unmerged implementation is not claimed to exist in the current RC/stable. Safety and correctness must not be inferred from PR closure. No changes to working version, pinned engine or published release.