[masarray]

P0 SG7 reporting: preserve true per-RCB setup write evidence, including zero-frame paths

Confirmed causal observability gap (existing operator evidence)

SG7_BCU was field-tested 77/77 and 4/4 routed RCBs on Oct 24–25, but 28 Oct recurrence showed 56/77 and 1/4, despite identical pinned engine and reporting source. The October diagnostic shows all four StartConfiguredStaticReportMonitor attempts returned success, but A01 alone had a received Report. Do not equate successful Start/RptEna/GI write with traffic proof.

Current ARSAS already captures ARIEC NativeReportMonitorStartResult.RptEnaWriteAccepted, GeneralInterrogationWriteAccepted, and ActivationReturnedAtUtc into the session's ReportActivationByPlanId. However ProcessReportHealth creates ReportStreams only after a routed frame. Thus the new merged #761 Copy Diagnostic incorrectly shows unverified write evidence for all zero-frame RCBs even when the engine had a concrete result.

Scope (6 files, no network behavior change)

• Typed immutable Iec61850RcbActivationEvidence (native write result, not readback or actual Report).
• One atomic read-only device.RcbActivationEvidence per current association, published only after a completed setup attempt, bounded to 64 plan IDs, cleared on start/stop/reconnect.
• Existing Copy Diagnostic joins RCB + DataSet exactly, accepts only one matching activation, and distinguishes activationCall=success/failed/unknown, setupRptEnaWrite, setupGiWrite, activationReturnedUtc from actual Report Frames, GI frame reason, and value completeness. Ambiguous/wrong-DataSet/old-association proofs remain unknown.
• Focused regression tests for SG7 A01 routed vs I01 zero-frame with accepted RptEna/failed GI, wrong DataSet, duplicate setup identity, and null reset.

Explicit non-goals / safety

No new IEC 61850 MMS reads/writes, no GI retransmit, no dynamic DataSet changes, no retry or cyclic polling, no receive scheduler or projection changes, no Command/Write UI changes, no ARIEC pin repin, no main/stable publishing, no claim that #707's 21 missing process values have been restored or that 4/4 RCBs were received in the failed session. Operator FAT evidence does not need to be re-captured to validate this diagnostic change.

Based exactly on integrated RC e7b1943792488af4a4dd03841c39abfe017f121a. Kept DRAFT until current-head Windows Build + app test pass and review.

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-10-28T00:01:53.650598Z	11365f5	Draft marked ready

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.