ClusterRoleInfo resolved the leader address through Cluster.State.Members, which is
published asynchronously; a miss stored null permanently, so RoleLeader(role) stayed null
(flaky test timed out under load: 6/10 failing with CPU contention, 0/20 after). Only the
address is ever read, so store it directly. Adds a deterministic unit test (fails without
the fix) and InternalsVisibleTo for the Cluster.Tests project.
Claude-Session: https://claude.ai/code/session_01Ef2ozxS7QPanGT8uoATbTe
The driver stamps DateTime.UtcNow at init and again per read pass; under CPU
contention both land in the same tick, so the strict greater-than assertions
flaked (reproduced 1/40 under load, sibling test A_cancelled_pass_still_flushes...
failed identically with equal timestamps). Test-only race; no production defect.
Claude-Session: https://claude.ai/code/session_01Ef2ozxS7QPanGT8uoATbTe
Review of 75e559f6:
- OnConditionReportEvent now mirrors MonitoredNode2's AuditEventState guard (only while Server.Auditing,
over SignAndEncrypt / https-uabinary). Conditions report their Part 9 method audit events through
themselves, so the first cut queued AuditConditionAcknowledge (i=8944) etc. to every i=2253 item with
auditing off -- the removed folder->Server forward used to drop them.
- ReconcileAllEventsMonitoring keeps one AreEventsMonitored unit on every materialised condition while any
i=2253 item exists (called from SubscribeToAllEvents, materialise, removal, rebuild), so
ConditionState.EventsMonitored() is true on a server whose only subscribers are Server-object items.
A protected SubscribeToEvents override no-ops all-events items, which stops the SDK's AddRootNotifier
(unsubscribe: true bug) taking that unit back from conditions rebuilt under a live item.
Flag-off over-the-wire coverage: Acknowledge/AddComment with and without a folder co-subscriber (no audit
event, transitions once each), EventsMonitored on/after-rebuild/off, rebuild survival with one copy per
event, native + scripted conditions once each, refresh once per condition with RefreshStart/End; plus an
ACL-on native-vs-scripted notifier-folder decision. Falsified: removing the audit guard reds both audit
cases; removing the reconcile reds the monitored test.
Claude-Session: https://claude.ai/code/session_01Ef2ozxS7QPanGT8uoATbTe
The libplctag-path positive control asserted 'timed out', but libplctag's internal
timeout equals the probe timeout and races the probe's own cancellation; under CPU
load it surfaces as 'ErrorTimeout' instead. Accept either libplctag stall outcome
(native never produces either: it fails at once with 'Connection lost').
Claude-Session: https://claude.ai/code/session_01Ef2ozxS7QPanGT8uoATbTe
Server-object (all-events) items bypassed the ACL entirely: the SDK attached them to every root
notifier's fan-out via SubscribeToAllEvents, and a root notifier with no direct subscriber forwarded
to the core Server object, which this node manager never sees.
- SubscribeToAllEvents override records the item and attaches it to nothing.
- Each condition's own OnReportEvent (fires once per event) queues the event to each recorded item
whose session holds Subscribe on the condition (RawPath) or any notifier folder it reports through
-- the per-folder path's question; otherwise mirrors MonitoredNode2.OnReportEvent.
- The root-notifier forward to the core Server object (OnReportEvent override) is dropped.
- ConditionRefresh on an i=2253 item is filtered the same way (once per condition).
- Revalidation is implicit: every event re-asks the authorizer (per-generation cache, key expiry and
revocation re-checked per call). Inert with Security:Acl:Enabled=false.
Over-the-wire tests: no-grant vs granted (positive control), created-before-deploy, revoke/restore by
deployment, key expiry with no tick, refresh, and flag-off. Falsified: disabling the filter, or
restoring the forward, turns the enforcement tests red.
Claude-Session: https://claude.ai/code/session_01Ef2ozxS7QPanGT8uoATbTe
Ab.Transport.Tests: predecessor-wait counter instead of 300 ms (DeviceSessionPoolTests x2), throttle
queue/interval-park hooks instead of 50/100 ms (ForwardOpenThrottleTests x2), UnobservedTaskProbe with a
must-report control fault instead of N GC cycles (EncapsulationClientTests), re-armed keep-alive timer as
the control for the never-opened keep-alive test.
AbCip.Tests: still-polling sibling / control driver read counts instead of 300-500 ms (subscriptions x4,
alarm projection x5), structural ProbeTask==null + enabled-probe control (host probe x2), predecessor
wait (NativeAbCipLeaseReleaseTests), processed waits in the shutdown/logging tests.
AbLegacy.Tests: sibling-subscription control (Unsubscribe_halts_polling), ProbeCts==null + control.
AbCip.IntegrationTests differential safety: a post-phase control open through the same throttle instead
of 1.5-2.5 s sleeps (via a native tag init, keeping the harness read-only guard green).
Fake tag ReadCount made interlocked; CapturingLogger entries snapshotted under a lock.
Claude-Session: https://claude.ai/code/session_01Ef2ozxS7QPanGT8uoATbTe
- Process-wide CipConnection statics (ForwardOpenCount, Forward Close counts, RequestsSent, exchange
stats, OpenAttempts, OpenAttemptBudget) move to an instance ConnectionDiagnostics owned by each
DeviceSessionPool (and by a stand-alone CipConnection). The open-attempt budget becomes a disposable
per-DeviceKey limit (LimitOpenAttempts); the differential harness scopes one to the press under test.
- LibplctagOrphanRelease.ForceRootingForTest is gone: ReleaseIfOrphaned(tag, canReadHandles) overload.
- A connected reply with general 0x01 / extended 0x0107 or 0x0203 (target no longer holds the class-3
connection, TCP still up) retires the session at once; the next send reopens through the throttle.
- New tests: dispose during RegisterSession, abandoned-open observer (falsified), scheduler-dispose-throws
guard (DeviceSessionPool.CloseAsync), release-in-background with a shared open in flight, per-device
limit scoping, CIP-close recycle (request + keep-alive + negative control).
Claude-Session: https://claude.ai/code/session_01Ef2ozxS7QPanGT8uoATbTe
#562: direct event (alarm) monitored items are now tracked in a second map
(_eventItems, fed by the SubscribeToEvents override) and re-authorized by
RevalidateMonitoredItems on both triggers (DataPlaneAuthorityState.Changed and
AclRevalidationTimer). A denied item is detached via the SDK's own protected
SubscribeToEvents(unsubscribe: true) - never MonitoringMode.Disabled - and
re-attached on restore; a client delete of a stood-down item is absorbed.
Over-the-wire tests: revoke -> stop (positive control on a still-granted
notifier) -> restore -> resume; clock-driven expiry via the real timer with a
control session; delete-while-stood-down is not resurrected; unexpired
negative control. Server-object (all-events) subscriptions are documented as
outside the ACL (observed over the wire).
#563: DataPlaneAuthorizer.AuthorizeKeyUsable runs the FindUsableKey probe on
the WriteOperate and AlarmAck role gates regardless of Security:Acl:Enabled,
so an expired or deployment-disabled key loses its activation-time roles on
its open session. Anonymous (no key id) is unaffected; reads/subscriptions
stay ungated with the flag off. No new flag.
Docs: docs/security.md and the CLAUDE.md API-key/ACL section.
Claude-Session: https://claude.ai/code/session_01Ef2ozxS7QPanGT8uoATbTe
#570
- STRING: the built-in 88-byte Logix STRING holds 82 characters (LEN + DATA[82] + 2 pad), not 84; the
write still sends the whole 88-byte structure and a read never decodes the pad as characters.
- Mistype churn: a type refusal (CIP 0xFF/0x2107) on the existence read no longer throws from
InitializeAsync, so the runtime is not disposed/re-created (and re-Forward-Opened) every poll; the driver
also keeps a runtime that throws TypeMismatch from init and records the device as reachable.
- Malformed tag name: a TagPath that does not parse, or a name the CIP symbolic encoder refuses (now also
non-ASCII, empty member, >255-char member, unterminated subscript), is AbCipTagConfigurationException ->
per-tag BadConfigurationError (named const in AbCipStatusMapper), warned once, never feeding the device
backoff or health; siblings keep reading. A malformed probe tag no longer feeds the backoff either.
- Micro800 BYTE/WORD/LWORD (0xD1/0xD2/0xD4) are type-checked as integers of their width: REAL/LREAL or a
narrower integer over LWORD is TypeMismatch, not decoded garbage.
#579
- ConnectionLease.ReleaseInBackground: one non-blocking release shape for both native AB runtimes. The
AbCip runtime's Dispose no longer waits up to 2 s under the per-tag lock; the pool still registers the
close synchronously so a re-create cannot overlap it. Offline soak tests (25 cycles, AbCip + AbLegacy)
prove no leaked connection; FakeEipServer gains OpenConnections.
Claude-Session: https://claude.ai/code/session_01Ef2ozxS7QPanGT8uoATbTe
#572: listing continues at max(instance in page)+1 (no ascending-order assumption, no-progress guard kept);
browsePageDelayMs (default 5 ms) paces pages; DiscoverAsync reads templates through one batch reader
(IAbCipTemplateReaderFactory.CreateBatch) holding one lease.
#575: differential harness reports a same-status template refusal as an equal refusal; offline test pins the
open-attempt budget without relying on the abort ending the phase.
Claude-Session: https://claude.ai/code/session_01Ef2ozxS7QPanGT8uoATbTe
CIP codec (#564): PatchU16 bounds guard; CipReply content equality; CipPath
port/length/ASCII/odd/oversize validation with ArgumentException for malformed
tag names and routes; CipPath.MessageRouter/ConnectionManager are now
ReadOnlySpan<byte>; refused Forward Open surfaces ForwardOpenRefusedException
with structured general/extended status; test CaptureFile uses BinaryPrimitives.
EncapsulationClient (#565): frame write bounded by the exchange timeout, which
faults (never cancels) the connection; finite positive timeouts enforced at
every entry point; gate wait bounded by the timeout; deterministic reorder of
the 300 ms test; new dispose-during-drain, gate-wait and RegisterSession
zero-context tests.
Claude-Session: https://claude.ai/code/session_01Ef2ozxS7QPanGT8uoATbTe
AbCip: node-level Drivers:AbCip:DefaultTransport and the working pin
recipe. AbLegacy: default libplctag, Native opt-in, exact wire evidence.
Type-mismatch and Unsupported are refusals that never evict or degrade.
Claude-Session: https://claude.ai/code/session_01Ef2ozxS7QPanGT8uoATbTe
HandleOf cast the reflected nativeTagHandle with (int) and was called
inside the runtimes' catch blocks, so a libplctag bump that retyped the
field would replace every create failure with an InvalidCastException.
The field is now accepted only if it is still an int (otherwise
CanReadHandles is false and orphans are rooted), and HandleOf never
throws. Both projects pin the field type explicitly.
Claude-Session: https://claude.ai/code/session_01Ef2ozxS7QPanGT8uoATbTe
A native TypeMismatch (read status, write gate, or CIP 0xFF/0x2107) used
to evict the runtime and degrade health on every poll; for a sole lease
holder that meant Forward Close + Forward Open per poll, and an all-
mismatched driver tripped the comms-loss watchdog. It now reports
BadTypeMismatch, keeps the handle, leaves health alone and warns once per
tag per instance — reads, whole-UDT reads and writes, plus the libplctag
width check. Tests (red against the pre-fix driver): single create over N
polls, unchanged health, one warning, and one Forward Open / zero Forward
Close over the real native runtime against FakeCipTarget. AbLegacy has no
TypeMismatch path (its native runtime has no type check).
Claude-Session: https://claude.ai/code/session_01Ef2ozxS7QPanGT8uoATbTe
ConfigureSharedLogger had no production caller, so the pool's only
diagnostics (first-wins options conflicts, scheduler/close failures)
went to NullLogger. DriverFactoryBootstrap now configures it from the
driver registry factory and from the AbCip/AbLegacy probe factories
(admin-only nodes reach the pool only through Test Connect). Wiring tests
reset the pool to a null logger first, so removing the call goes red
(verified).
Claude-Session: https://claude.ai/code/session_01Ef2ozxS7QPanGT8uoATbTe
The old AdminUI cannot author the per-driver transport key (#576), so the
'pin before upgrading' recipe had no working path. A node-level appsetting
(env Drivers__AbCip__DefaultTransport) now supplies the default for AbCip
drivers with no key: explicit key > node default > compiled Native. Bound
and validated unconditionally (bad value fails host start); the driver
factory and the Test Connect probe both read it so they agree. AbCip.md
upgrade recipe rewritten around the key; Configuration.md documents it.
Also documents first-wins for AllowPacking/probe and host-alias throttles
(final review M6) and moves the status-bit note out of AbCip.md (M5).
Claude-Session: https://claude.ai/code/session_01Ef2ozxS7QPanGT8uoATbTe
On a wire the native PCCC path is proven only for N-file reads/writes and
one F8 read (ab_server). An absent transport key now resolves to Libplctag
for every family; explicit Native still wins and still warns once per
instance on PLC-5/LogixPccc. AdminUI dropdown, CLI help, evidence claims
(PcccFileType, AbLegacy.md wire-evidence table, release note) corrected:
S:0 (default probe) and B/T/C/R/S/L/ST/A/O/I are unverified natively.
Claude-Session: https://claude.ai/code/session_01Ef2ozxS7QPanGT8uoATbTe
Native is the default, so a locally refused write must not look like a comms fault.
- AbLegacy: an AbTransportException with AbStatus.Unsupported (status-bit sub-element writes, PLC-5
non-N reads, all PLC-5 writes; libplctag's ErrorUnsupported maps here too) now returns
BadNotSupported on both the read and the write path (incl. the bit-RMW path). The tag handle is
not evicted and driver health is not touched. A native runtime whose init read is refused locally
is CACHED rather than disposed, so a poll of a refused tag does not re-create (and re-lease the
pooled connection) every cycle.
- AbCip: the same mapping on the write path and both read paths (single + whole-UDT).
- Uses the existing named BadNotSupported constants (StatusCodeParityTests guard).
- Tests (falsified by disabling the mapping, and separately the keep-on-refusal): status-bit
writes and PLC-5 read/write through the driver over the native runtime — BadNotSupported, zero
wire requests, one create, health unchanged; AbCip refused read/write via a refusing fake plus a
control that a non-refusal failure still evicts and degrades.
- Docs/CLAUDE.md: BadNotSupported; #571 known-gap wording removed.
Claude-Session: https://claude.ai/code/session_01Ef2ozxS7QPanGT8uoATbTe
- I1: the upgrade restart itself moves every no-key AbCip/AbLegacy driver to Native (every process
start constructs drivers, incl. boot-from-cache). Docs + CLAUDE.md now say so and tell operators
to set "transport": "Libplctag" BEFORE upgrading (older builds ignore the unknown key).
- I2: AbLegacy release note leads with the status-bit write refusal (T4:0.DN, C5:0.CU, R6:0.EN, ...)
on defaulted SLC/MicroLogix drivers and the fallback. The real client status is
BadCommunicationError, not BadNotSupported (the driver's write catch-all maps the runtime's
AbTransportException like any transport failure, as #571 does for PLC-5), with handle eviction
and Degraded health; pinned by a driver-level test (no 0xAA write on the wire, with a positive
control). ST string byte order stated as unverified. Also corrects the docs' "T4:0/DN" examples:
the driver's address grammar is "T4:0.DN" ("/DN" is a malformed address).
- I3: DriverProbeHandshakeE2eTests' two AbCip probe tests are a [Theory] over Native and
Libplctag. Run against ab_server controllogix: all 4 rows pass with identical exact messages.
- M1: documented (docs + resolver XML doc): the AdminUI merges only the probed device, so a
mixed-family defaulted AbLegacy driver can probe Native while running Libplctag.
- M2: factory ParseEnum rejects undefined integer strings ("7") via Enum.IsDefined (both drivers).
- M3: AbCip driver-level bad value reads "AB CIP config in '<id>' has unknown Transport ..."; the
named constant is gone (tagName: null now means driver-level).
- M4: the explicit-Native limited-family Warning is once per instance, not per (re)init.
- Nit: EffectiveTransport docs note that injected factories bypass it.
Claude-Session: https://claude.ai/code/session_01Ef2ozxS7QPanGT8uoATbTe
Task 22 (spec section 8 P5), with the controller rulings that exclude PLC-5 and LogixPccc.
- Transport is nullable on both option records; null (key absent) means "use the default" and is
never serialized. Factory DTO parse: absent -> null; a bad value still fails via ParseEnum.
- One resolver per driver (AbCipTransportResolver / AbLegacyTransportResolver), used by the driver
constructor, the Test Connect probe and (through the driver) the CLI.
AbCip: null -> Native. AbLegacy: null -> Native unless any device is Plc5 or LogixPccc, then the
whole driver resolves to Libplctag. Explicit values always win; explicit Native with a
Plc5/LogixPccc device logs one Warning at init.
- Init log shows configured vs effective, e.g. transport=Native (default).
- AdminUI forms: "Default (...)" / "Native (managed)" / "libplctag (fallback)". Default omits the
key on save (reverses Task 19's explicit re-emit, which would have pinned every edited driver).
- CLIs: --transport default|native|libplctag, default "default" via the resolver, so
--plc-type Plc5 with no --transport runs libplctag.
- Tests pinned to Libplctag where a default-path runtime was created: found by instrumenting the
native factories with the current xUnit test name (one unit test, AbLegacyReinitConfigAdoption;
plus the two fixture-gated AbCip probe E2E tests whose exact messages were proven on libplctag).
TransportMatrix and the differential harness already pass explicit transports.
- Switch/resolver/CLI/form tests are falsified by mutating the resolver (default -> Libplctag,
exclusion removed, AbLegacy default -> Libplctag): each mutation turns them red.
- Replaces the inline role-name literal in AbCipDriverFactoryExtensions' ParseEnum call with a named
constant; RoleAuthorityGuardTests was already red on the base commit for that line.
- Docs: AbCip.md / AbLegacy.md default + release note; CLAUDE.md AB section.
Claude-Session: https://claude.ai/code/session_01Ef2ozxS7QPanGT8uoATbTe
Pass 2 (2026-10-08 00:21Z, df776170): 17/17. Browse 3162=3162 tags, 20 UDT
templates identical, 200 tags read with 0 unexpected divergences (6 live
values compared structurally + every still element), 1 Forward Open and 1
acked Forward Close per native phase. Latency native p95 105.8 ms vs
libplctag 139.5 ms (limit 153.4) PASS. Pass 1 failures and their fixes are
recorded in the gate-results doc; pass 2 had no independent pcap.
Claude-Session: https://claude.ai/code/session_01Ef2ozxS7QPanGT8uoATbTe
The 5069-L3x supports the Large open, as the press proved. The narrow/refusing case is a pre-v20 1769-L3x or a bridge. AbCipDeviceOptions.ConnectionSize: the native transport honours it on the wire; libplctag ignores it.
Claude-Session: https://claude.ai/code/session_01Ef2ozxS7QPanGT8uoATbTe
M2: after plc_tag_destroy, the release writes 0 back to the wrapper's nativeTagHandle, so a repeat release is a no-op instead of touching a destroyed id. The runtimes keep the destroyed handle for tests and diagnostics.
M1: the rooting fallback is now described accurately, as an unbounded leak (one native tag per failed create), not a bounded one. A test seam (ForceRootingForTest) pins that the fallback roots the tag and destroys nothing.
Both pins are red with their line removed. The orphan tests share a non-parallel collection, because the counters and the seam are process-wide.
Claude-Session: https://claude.ai/code/session_01Ef2ozxS7QPanGT8uoATbTe
I1: a live verdict no longer waives the value check. Only elements that moved between the two libplctag reads are excused.
- An element equal in both libplctag reads must also be equal through native; otherwise **NO: stable element i**.
- A moved element must not read NaN or infinity on native unless libplctag saw it too; otherwise **NO: non-finite element i**.
- Live tags log their full raw bytes and every element, and the summary reports live tags as N of the total.
- The offline pin adds the pass-1 Real[5] shape: a stable element shifted on native fails. It is red with the element check disabled.
M9: after each native phase (differential and latency), wait for the session close. With AB_DIFF_COUNT_OPENS=1, assert exactly one Forward Close was sent, using the new CipConnection counters; acked is logged.
M8: each latency pass logs its GC collection deltas (gen0/1/2) for both transports, plus native per-exchange round-trip min/mean/max.
M4: the latency test also asserts that no libplctag orphan had to be rooted.
Claude-Session: https://claude.ai/code/session_01Ef2ozxS7QPanGT8uoATbTe
Joseph Doherty