The mxaccessgw OnWriteComplete correlation is live (server b948e69, worker
53f69cd — all four unary write kinds), so every write reply's statuses[0] now
carries the real COM commit outcome. Live-gating the cutover on windev found
and fixed three defects:
- StatusCodeMap.FromMxStatus truncated the correlated failure row's MxStatus
detail code to a byte (1007 & 0xFF = 239 -> OPC DA Good band), mapping a
REFUSED write to Good. New invariant: a row that is not IsSuccess() never
maps Good — details > 0xFF (and contradictory Good-band substatuses) route
by category (SECURITY_ERROR -> BadUserAccessDenied, named const covered by
StatusCodeParityTests; COMMUNICATION/CONFIGURATION keep their codes).
- The borrow-from-subscription-registry write optimization could never commit:
a subscription handle is plain-advised, and MXAccess refuses both the
no-login write through it (SECURITY_ERROR detail 1008, live-captured) and
the supervisory advise upgrade (E_INVALIDARG) — one advise per HANDLE, not
per attribute. Removed: writer subscribedHandleSource ctor param + borrow
branch, GalaxyDriver wiring, SubscriptionRegistry.TryResolveItemHandle +
its forward index and test file. The writer always AddItems its own
supervisory-advised handle, which commits fine while the same attribute is
subscribed (live-proven; pinned by the new write-while-subscribed smoke).
- (gateway-side, fixed there) the correlation initially covered WriteSecured
only, while the dominant no-login path emits plain Write.
Live gate, all legs PASSED against windev: happy write Good off a confirmed
status row with galaxy.writes.unconfirmed + advise_failed both FLAT; read-only
refusal surfaces Bad-family (the #5 node-revert input Galaxy could never
produce before); WriteCompletionWaitMilliseconds=0 control -> provisional Good
+ metered unconfirmed (override removed after, restoration re-verified). New
env-gated smokes: Live_correlated_write_statuses_confirm_commit_and_surface_refusal,
Live_write_to_subscribed_tag_commits_via_own_handle (replaces the borrow
smoke), GatewayGalaxyWriteMeterLiveTests (run filtered — meters are
process-wide).
The connect path was the one capability call the resilience pipeline never
saw, so a driver that could not connect contributed zero breaker input and
the /hosts resilience panel read 'All targets healthy' against a dead
endpoint. InitializeAsync now routes through the invoker as
DriverCapability.Initialize, keyed on the driver-instance id.
- No pipeline retry: the actor's RetryConnect timer is the retry loop
(parser forces retryCount overrides to 0, same funnel as write-shaped).
- Breaker threshold 2, not the Read-sized 5: the breaker samples 30 s and
the reconnect timer fires every 10 s, so at most three attempts fit one
window — threshold 5 could never open. Pinned by its own test.
- An open breaker fast-fails into the same InitializeFailed path as a real
connect failure; the 15 s break window costs at most one extra tick.
The best-effort teardown ShutdownAsync stays unwrapped.
- Companion fix: DriverCapabilityInvokerFactory.Create now invalidates the
instance's tracker snapshots with its pipelines — a respawn during an
outage discarded the only breaker whose OnClosed could clear the OPEN
flag, stranding a false 'Breaker OPEN' chip forever.
Live gate PASSED on docker-dev (image symbol-verified per container):
rtu-gate against dead 10.100.0.35:5021 opened its Initialize breaker on
both centrals with the red 'BREAKER OPEN' chip on /hosts; a socat
5021→5020 forwarder recovered it (both breakers closed within 1 s, panel
back to 'All targets healthy'); healthy MAIN-modbus recorded zero breaker
events. Full suite 50/50 green. Evidence:
docs/plans/2026-08-07-adminui-followups-and-live-gates.md §7.
The two items deferment.md §0.1 named as the remaining untracked engineering.
adminui-followups (19 pending / 0 completed) — CLOSED by the overlap audit the plan
itself asks for first, with no code written. All 19 tasks are shipped, superseded by
v3, or moot. WS1's premise had also expired: it assumed driver-embedded tag lists in
DriverConfig are the runtime source of truth, which v3 ended (the artifact populates
RawTags). Its capability shipped in the v3 shape — TagConfigEditorMap + DeviceFormMap,
both verified reachable from RawTagModal/DeviceModal rather than merely present.
Live gates on docker-dev (image rebuilt: the rig was running a build 3 h older than the
last code commit; parity asserted per container plus a positive check that the symbols
under test are in the shipped DLLs):
- R2-03 T11 PASSED — build 0 errors, Runtime.Tests 555/555, RoslynVirtualTagEvaluator 20/20.
- R2-03 T12 PASSED — VT 42 -> 0x80020000 -> 42, staged WITHOUT a deployment (a Modbus
register flipped on the fixture) so the VT child survived; a deploy-driven trip creates
a fresh child with no last value, which is correct-by-design and masked the result twice.
P7 confirmed: two byte-identical redeploys, zero disturbance.
- driver-reconfigure Task 3 PASSED — a faulted driver adopted a corrected config and
connected with the node's startedAt unchanged and RestartCount=0.
- R2-10 T10 still OPEN, but the recorded cause was wrong. Not ConnectionBackoff
throttling (the breaker opens on 2 failures / 30 s): the dead-endpoint failure mode
never enters the resilience pipeline. IDriver.InitializeAsync is not invoker-wrapped
(DriverInstanceActor.cs:707) and ModbusDriver swallows per-tag failures in BOTH
directions, so even the plan's fallback trigger cannot work. Live-proven: a driver
timing out every read against a blackhole reported ConsecutiveFailures=0 while the
panel read "All targets healthy."
Finding from T12: the Bad publish's documented "carrying the last-known value" is
unobservable by any client — subscription and a raw Read service call both return NULL.
The actor is correct; every existing failure assertion checked StatusCode alone, so the
value half was asserted nowhere. Adds
VirtualTagActorTests.The_Bad_result_carries_the_last_known_value, which proves the actor
emits it. The documentation is what overstates.
Records the outcome in the design doc §6 (including the stale-open race the
review found and fixed in both modals), closes the follow-up in the tracking
doc and the universal-browser plan, and registers RawBrowseModal's shared
race shape in deferment.md §0.1 as the deliberate leftover.
The register said "14 issues remain open" and recorded #491 as an open live
gate with "no gate-passed commit exists". Both were false: the tracker has
been at 0 open since 2026-08-04, and #491's gate passed that morning after
finding a production-fatal defect the "code complete" reading had missed.
Re-derived every count from its source of truth rather than from this file:
the Gitea API (0 open; closure dates for all 14), a parse of every
docs/plans/*.tasks.json (31 open tasks across 6 files), and a listing of
code-reviews/ (10 unreviewed projects, 3 retained Wonderware modules).
That surfaced three things reading the file could not:
- §6.2 was missing a live plan file that §5 lists as an open gate, so the
two sections disagreed on whether the work existed.
- §5 cited that file at a date that does not exist in docs/plans/.
- The gate's stated blocker has expired — it needed an operator sign-in,
and docker-dev has since disabled AdminUI login, so it is runnable now.
Also records a correction rather than a status flip: archreview 03/S4
("primary gate default-allows on unknown role") is the same shape as a
decision deliberately taken in PrimaryGatePolicy.ShouldRecordHistorizedValues,
where an unknown role captures. Defensible there, not in a write gate — so
closing 03/S4 must not generalise one rule across both.
Third round running in which this file's own bookkeeping was what broke,
after the dangling pre-amend SHAs and the truncation. The trailer now says
so, and says to re-derive rather than re-read.
Claude-Session: https://claude.ai/code/session_015p7wGqy3YpZNCpDzTpGMKo
#519 — a driver's OPC UA status code now reaches a client with its sub-code intact.
DriverInstanceActor projected the driver's uint onto a 3-state Good/Uncertain/Bad enum
and OtOpcUaNodeManager re-expanded Bad to a bare 0x80000000, so BadNoCommunication,
BadTypeMismatch and BadNotSupported were indistinguishable on the wire and every
driver's status constants (incl. the 16 #497 had just corrected) were diagnostic in
logs only. Resolved by option (b): the raw uint rides AttributeValuePublished ->
AttributeValueUpdate -> IOpcUaAddressSpaceSink.WriteValue, OpcUaQuality is deleted, and
severity classification is a helper (OpcUaStatus.IsGood/IsUncertain/IsBad).
The surface was wider than the issue costed (41 files, not "the sink and 3
implementations"): OpcUaQuality had non-driver producers with no uint to carry.
Each got its natural code — BadInternalError for a failed virtual-tag script,
BadNoCommunication for driver comms loss — and ScriptedAlarmHostActor's pair of
mutually-inverse uint<->enum mappers is deleted, having round-tripped a code the
engine already held. AddressSpaceApplier's removal writes deliberately keep the
generic Bad: not losing a code the driver had is this issue; inventing one for the
deploy path is a separate call.
Behaviour change: WriteAlarmQuality fires its Part 9 event on a status-code change
rather than a severity-bucket change. While the annotation was 3-state those were the
same predicate; now a move between two Bad reasons changes what a subscriber reads.
New SDK-free named constants live on Commons.OpcUa.OpcUaStatus, and
StatusCodeParityTests is widened to cover them — that guard is reflection-driven, so
it can only see a named constant, never an inline literal.
#517 — the symptom is real but the issue named the wrong path and its fix was unsafe.
ShutdownAsync cannot race: DriverInstanceActor calls it from PostStop, after the
mailbox drains. (Core/Hosting/DriverHost, the one class with a concurrent GetHealth,
has no production call site.) InitializeAsync can: it runs on a fire-and-forget
Task.Run, and MTConnectDriver writes Healthy then starts loops that can throw into a
catch which disposes the client before writing Faulted.
Moving the write earlier, as suggested, inverts the terminal state — teardown drains
in-flight loops that write health as they unwind, so a late loop could leave Healthy at
rest with every connection closed. The fix is a bracket (conservative before, terminal
after), applied to MTConnect's init-failure path only; shutdown keeps its single
trailing write, and the rule plus the reachability table are documented on
IDriver.ShutdownAsync.
CalculationDriver.GetHealth returned a hard-coded Healthy that consulted nothing —
Healthy before initialize, after shutdown and after dispose. Now derived from _running.
Its test, named Health_is_always_Connected, asserted the defect and is replaced.
Every new guard was falsified by breaking the production code and watching it go red.
The two sub-code tests each assert two different Bad codes, since a single-code
assertion could pass against a hard-coded constant under the old projection.
Pre-existing failures, verified unchanged on a stashed tree and not chased: 4 in
Driver.AbLegacy.IntegrationTests (slc500/micrologix/plc5 fixtures not up) and 3 in
Host.IntegrationTests.
Closes#517Closes#519
Claude-Session: https://claude.ai/code/session_015p7wGqy3YpZNCpDzTpGMKo
Records the correction rather than quietly re-scoping: the delete recommendation for #525 rested
on "1 of 12 flushes actually frees memory", derived by grepping for two specific cache-clear
calls and treating absence elsewhere as proof. Real figure is 3 of 12, and MTConnect's is a
designed, tested path. The sweep (12 drivers + IDriver + ~15 test files) was reverted rather than
finished on a justification that had collapsed.
Claude-Session: https://claude.ai/code/session_015p7wGqy3YpZNCpDzTpGMKo
Also corrects two claims this register itself got wrong, which changed both decisions:
- §3.2 said DriverHostStatus was "deliberately re-created in the v3 initial migration, so
deleting on inference would be wrong". The publisher it waited for cannot be written at
all — Phase 4 leaves a driver-only node with no ConfigDb. "Deliberately re-created" meant
carried forward mechanically.
- §3.3 said the Tier-C-only protections were dormant, implying Tier A/B ones ran. Nothing
constructed MemoryTracking either, so the whole Core/Stability layer was dead — which is
why 'pass real tiers' was never a flag flip.
Open count stays 17 (verified against the tracker API, not derived — the derivation was wrong
last round) but membership changed: #521/#522 closed, #524/#525 opened.
Claude-Session: https://claude.ai/code/session_015p7wGqy3YpZNCpDzTpGMKo
The enumeration already listed the post-#523 set (17 items) while the count said
18 -- the number was written between closing #516/#518/#507 and closing #523.
Verified against the API rather than re-derived: 17 open.
Claude-Session: https://claude.ai/code/session_015p7wGqy3YpZNCpDzTpGMKo
Closed#516, #518, #507 and #523, each with a resolution comment. The comment
matters more than the state change in three of the four cases, because what was
actually done differs from what the issue asked for:
#516 — the issue's own survey was INCOMPLETE. It named Modbus/FOCAS/OpcUaClient;
the defect also hit AbLegacy (missed) and Sql (deliberate, on a false
premise). The load-bearing fix was the spawn seam, not any driver.
#518 — only HALF resolved. IRediscoverable is consumed as an operator prompt,
never an address-space rebuild; IHostConnectivityProbe is untouched and
split to #521. Closing without saying so would have implied both.
#507 — SUPERSEDED, not migrated. The requested migration was impossible: both
of the path's inputs are structurally empty in v3, so lifting the guard
would have injected nothing.
#523 — done, carrying forward that AbCip and FOCAS now have no tag-change
signal at all.
#523 was not in the set named for closing, but it had become resolved-but-open in
the same session by the work above, which is the same category; closing it is
noted here rather than done silently.
Register reconciled: the "three issues resolved but still open" note is replaced
by a table of what each closing comment records, and the two issue counts that
said "21 open, of which 3 are stale-open" now say 18 open, none stale.
Claude-Session: https://claude.ai/code/session_015p7wGqy3YpZNCpDzTpGMKo
f3cd685d deleted all 941 lines of deferment.md. The file was empty through the
merge to master and was only noticed when a follow-up edit's grep matched
nothing. Restored from 4624141a; diff vs. that version is +52/-3, all intended.
CAUSE, a Python evaluation-order trap rather than a merge conflict. The edit
script ended with:
open(p, 'w').write(open(p).read().replace(a2, b2))
Python evaluates `open(p, 'w').write` before its argument, so the file was
TRUNCATED before `open(p).read()` ran -- that read returned "", and "" was
written back. The assert guarding the edit had already passed against the
in-memory copy, so nothing raised.
WHY IT SURVIVED TO A MERGE. Every check I ran could pass on an empty file:
dotnet build cannot notice an empty Markdown file, the test suite does not read
it, and `git status --short` showed `M deferment.md` -- a modification, exactly
what I expected. Nothing asserted the file still had content.
Also re-applies the two edits f3cd685d was supposed to make (the §7.2 missing-
driver-docs row and the §7.2 status banner), records merge 2a0461a1 in §9 with a
round-2 item table, and adds a §9 entry describing this incident -- including
that it is the SECOND time in two rounds that the register's own bookkeeping step
damaged the register. Both failures share a shape: the verification step and the
mutation step were the same action, so a broken mutation produced a clean-looking
verification.
Claude-Session: https://claude.ai/code/session_015p7wGqy3YpZNCpDzTpGMKo
Closes the §7.2 priority-3 item. Sql was a shipped, merged driver with NO
user-facing documentation at all -- only design and plan files, which describe
what was intended rather than what an operator needs to author one.
Sql.md covers: SQL Server as the only constructed provider (the other four
SqlProvider names are reserved, and authoring one fails rather than falling
back), the KeyValue / WideRow tag models with worked TagConfig, Query being
rejected end-to-end, the config table with real defaults read from
SqlDriverOptions, connectionStringRef never carrying credentials into an
artifact, per-source query grouping and its case-sensitive group keys, and the
value-bound-vs-identifier-validated split that makes it injection-safe.
Calculation.md covers: what a pseudo-driver is and why it occupies a driver slot,
a Calculation-vs-VirtualTags decision table (the two are easy to confuse), the
literal-only ctx.GetTag dependency extraction and the runtime-built-path trap,
the two deploy gates, the all-deps-arrived publish gate, and the Good->Bad-once
error semantics.
Two stale claims found by checking against source rather than inheriting them:
- The driver README's Calculation row still carried G-1 ("no typed config form,
RunTimeout unauthorable"), fixed on 2026-07-27 by CalculationDriverForm.
- docs/Raw.md's "one auto-created default Engine device" NEVER SHIPPED. I had
copied it into the new doc before checking; "Engine" has zero occurrences in
src/ and nothing special-cases Calculation at device creation. Corrected in
both files. The config key was also wrong in my first draft: runTimeoutMs, an
int in milliseconds, not a TimeSpan string -- and it is ignored on reinit
because the evaluator's timeout is constructor-fixed.
Full solution: 49 suites pass; the 2 failures are the known pre-existing
environment ones (AbLegacy.IntegrationTests docker fixture, Host.IntegrationTests).
OpcUaClient.IntegrationTests, which was silently failing, is now green.
Claude-Session: https://claude.ai/code/session_015p7wGqy3YpZNCpDzTpGMKo
tag gate, and fix three silently-broken OpcUaClient integration tests
Continues working through deferment.md's remaining items.
SKIPPED TESTS (13 -> 0 for this reason). The skip said "dark until Batch 4";
Batch 4 shipped as v3.0 and RETIRED the equipment-tag path rather than lighting
it, so every one of them was waiting on something that had already happened and
gone the other way. Resolved individually rather than in bulk:
- REVIVED onto the raw/UNS shape (6): four DriverHostActor primary-gate cases,
the cluster-scoping rebuild, and the real-SDK dual-namespace materialisation
E2E. The behaviour never went dark, only the v2 seeding did.
- FOLDED into a live parity test (3): DeploymentArtifactRawUnsParityTests
already compared RawTagPlan record-equal and RawTagPlan carries array /
historize / alarm intent, so widening its corpus by two tags covers what
three empty placeholder suites had been promising. Those suites are deleted.
- DELETED (4): subjects genuinely retired -- equipment-namespace EquipmentTags,
and both dot-joint {{equip}}.X token suites (that resolver was deleted in
Batch 3; the slash-joint form is covered by DeploymentArtifactEquipRefParityTests).
Falsified, not assumed: removing the seeded UnsTagReference turns two revived
primary-gate tests red; re-homing the SITE-A node to MAIN turns the scoping test
red. Widening a parity corpus also needed direct value assertions -- parity alone
cannot distinguish "both seams right" from "both seams blind".
Runtime.Tests skips 13->3, OpcUaServer.Tests 4->1; all remaining are deliberate
EquipmentDeviceBindingRetired tombstones.
G-7 (deploy-time TagConfig gate). MQTT shipped an Inspect() nobody ever called;
now wired. Replaced the hand-maintained list with a reflection guard, because the
existing test enumerates driver types by hand and so guards renames but can never
notice a gap. NOTE: the first version of that guard was hollow -- it discovered
candidates via GetReferencedAssemblies(), which is circular, since the compiler
elides a reference nothing in the IL uses. Removing MQTT from the map also removed
its Contracts assembly from the manifest, and the guard passed green with the bug
reintroduced. Rewritten to scan the output directory; re-falsified and it now
fails naming MqttTagDefinitionFactory. Residual recorded at the code: Sql,
MTConnect, Calculation and OpcUaClient have no Contracts-side Inspect at all, so
the deploy gate stays weaker than the AdminUI's authoring gate for them.
OPCUACLIENT INTEGRATION TESTS (3 of 4 failing, on master too). Not fixture rot:
the simulator was healthy and Client.CLI read the very same node Good. v3 made the
driver resolve every reference through its authored RawTags table, so a bare
"ns=3;s=StepUp" fails LOCALLY at the resolver with BadNodeIdInvalid and never
reaches the wire -- the tests were exercising the unresolved-reference guard.
Fixed by seeding OpcPlcProfile.RawTags in the deploy artifact's TagConfig shape
and addressing by RawPath. 4/4 pass.
DOCS. Applied the Phase7* -> AddressSpace* rename across the four live docs
(historical bannered files deliberately untouched), resolving the three that are
not renames to their real members. Found en route that the earlier banner pass
missed three files: VirtualTags.md, OpcUaServer.md and ScriptedAlarms.md all
still presented the test-scaffolding GenericDriverNodeManager as the production
dispatch path. All three now carry the correction.
Claude-Session: https://claude.ai/code/session_015p7wGqy3YpZNCpDzTpGMKo
Removes the tag-set drift detector shipped in e77c8a35. The design was sound
-- the tautology drivers were correctly gated out, and the structurally
undetectable "vanished" direction was declared rather than faked -- but none of
that is the objection. The feature browsed real plant equipment on a recurring
timer, at a frequency no operator could configure (the interval was a
constructor parameter with no appsettings key), and that outbound traffic was
never once observed against a real PLC, CNC or MTConnect Agent.
Removed: TagSetDrift/TagSetDriftDetector; DriverInstanceActor's drift timer,
tick message, Props/ctor parameter, gate, handler and leaf-collector;
ITagDiscovery.AuthoredDiscoveryRefs + .DiscoveryStreamIncludesAuthoredTags and
their four implementations; 14 tests.
Kept: ITagDiscovery.SupportsOnlineDiscovery (it predates this and drives the
universal browse picker), and the whole IRediscoverable consumption from
09a401b8 -- the driver tells us, we do not poll.
CONSEQUENCE, deliberately accepted: AbCip and FOCAS browse a live backend but
implement no IRediscoverable, so they now have NO tag-change signal at all -- a
PLC re-download is invisible until someone re-browses by hand. Recorded in
CLAUDE.md so the next reader does not have to rediscover it, along with the
shape to reach for if that coverage is wanted back (event-driven, or triggered
by a reconnect/deploy, not a wall-clock timer).
Runtime.Tests 476 pass / 13 skipped (skip set unchanged); FOCAS 275, TwinCAT
192, MTConnect 491, AbCip 342, Core.Abstractions 266 all green.
Claude-Session: https://claude.ai/code/session_015p7wGqy3YpZNCpDzTpGMKo
The register recorded its remediation in §9 but left §1-§8 reading as the
original audit, in present tense. A document whose purpose is catching drift
had started drifting: it said "17 open issues, all confirmed still open" when
three were resolved and four new ones existed, and it described three dead
seams as live when one had been fixed and the other two documented and tracked.
Rather than rewrite the audit — which is worth keeping as the record of what
was found — §1-§8 are now explicitly framed as "as at 90bdaa44" and each closed
item is marked in place, so a reader can see both the finding and its outcome.
New §0.1 states what is actually left as at f0bd73bc.
Corrections where the audit was wrong, now visible at the point of the claim:
- §4.4 called the 18 DiscoveryInjectionDormantV3 tests "Real — blocked on
#507". They were v2 characterization tests asserting a graft v3 cannot
produce, so they could never have unskipped; deleted.
- §3.1/§7 named ReadWriteOperations.md as the highest-risk doc. security.md was
worse — it claimed anonymous sessions are default-denied when they can read
the whole address space.
- §3.3 missed that driver-stability.md's separate-Windows-service Tier C hosting
no longer exists either.
Also records three issues that are resolved but still open on the tracker
(#516, #518, #507), the four new ones (#520-#523), and adds the un-gated drift
detector to §5 — a gate that #523 would close by deleting the feature.
Footer now states that where §1-§8 and §9 disagree, §9 is current, and that
some §1-§8 path:line references no longer resolve because the code they point
at was deleted.
Three corrections to deferment.md, plus a CLAUDE.md warning.
The commit column in §9 was wrong for every stage. Each SHA was captured with
git rev-parse HEAD and then written into the file by amending that same commit
— which changes the SHA — so every recorded value was the pre-amend one: a
dangling object on no branch, reachable only through the reflog and due to
vanish at the next gc. The register was citing commits that effectively did not
exist. Now verified against git log e08855fb..914d70bf.
§9 also still described this as work on a feature branch; it merged to master
as 914d70bf.
The 2026-07-28 drift-detector entry now carries a SUPERSEDED banner. The
feature is in master and the entry stays as the record of what was built and
why, but it is not the intended end state: the owner's decision is to remove
the periodic re-browse (#523). The objection is the one the original entry
never weighed — the check is recurring, unconfigurable and never live-gated
outbound traffic to real devices, and how often you browse someone's PLC is an
operational decision rather than an implementation detail.
Both the banner and #523 record the consequence rather than arguing it: AbCip
and FOCAS implement no IRediscoverable, so after removal they have no
tag-change signal at all. TwinCAT and MTConnect keep their native event-driven
signals and lose only redundant polling. The IRediscoverable consumption itself
is explicitly out of scope for the removal.
The second thing #516/§8.2 deliberately did not build. The original objection
stands — Modbus, S7, MQTT, AbLegacy and Sql echo authored config from
DiscoverAsync, so diffing discovered against authored is a tautology for them
— but it is now ENCODED rather than used as a reason not to build.
The discriminator already existed: ITagDiscovery.SupportsOnlineDiscovery is
documented as "enumerates the tag set from the live backend rather than
replaying pre-declared/authored tags", and it separates the fleet cleanly —
FOCAS, TwinCAT, MTConnect and AbCip true; the five config-echo drivers
explicitly false. The check runs only for a driver that is
SupportsOnlineDiscovery AND reports the new AuthoredDiscoveryRefs. That is
nullable rather than empty-by-default on purpose: an empty collection
legitimately means "nothing authored, everything discovered is new", so
conflating the two would report total drift for a driver that never opted in.
Where the value is: AbCip and FOCAS browse a live backend but implement no
IRediscoverable, so before this they had no change signal at all. A periodic
compare is the only way to notice a PLC re-download.
A finding that changed the design: FOCAS, TwinCAT and AbCip all re-emit their
authored tags into the same DiscoverAsync stream as device-derived ones, so
the browse picker can show an operator their existing tags. Discovered is
therefore always a superset of authored, and a VANISHED tag can never appear
missing — one whole direction is structurally undetectable for three of the
four. Shipping a Vanished list that is permanently empty for them would have
been the half-inert seam this whole exercise removes. So the driver declares
DiscoveryStreamIncludesAuthoredTags, the detector does not compute the
undetectable half, and the operator message says "missing-tag detection
unavailable for this driver" rather than letting the absence of a missing-tag
clause read as reassurance. MTConnect is the only driver where both directions
work — its stream is purely probe-model-derived.
Behaviour: a failed browse is not drift (an unreachable device would otherwise
report every authored tag as vanished); a truncated capture is skipped for the
same reason; an unchanged drift is reported once rather than re-stamping its
timestamp; drift that resolves clears the prompt. Interval defaults to 5
minutes — it browses a real device, and a tag set changing is an engineering
event, not a runtime one. It reuses the Stage-2 surfacing, raising the same
/hosts re-browse prompt, and never rebuilds the served address space.
Comparison logic is a pure, separately-tested type rather than actor-inline.
14 new tests. The gate was verified load-bearing by deleting the
SupportsOnlineDiscovery half: the tautology-driver test goes red, and it
asserts discovery was never invoked rather than merely "no prompt raised",
which would have passed for the wrong reason if the timer never fired.
Full suite green except the same three pre-existing fixture-gated integration
suites, identical counts.
The first #516 pass deliberately left these two out, because each derives more
than options from config: Sql its ISqlDialect and resolved connection string,
FOCAS the IFocasClientFactory its Backend key selects — all injected at
construction. Adopting new options alone would poll a NEW tag set through the
OLD database/backend, and a half-applied config change is worse than a
discarded one because it looks like it worked.
Rather than accept that exclusion, this fixes the blocker. Each factory now
exposes a ParseBinding returning EVERY config-derived dependency as one value,
and the driver adopts them atomically:
- Sql: ApplyBinding takes (options, dialect, connectionString) and rebuilds
everything downstream — provider factory, Endpoint, and the SqlPollReader
that captures all three. It runs BEFORE BuildTagTable, so the tag table and
the connection it is polled over always come from the same revision. A
test-injected DbProviderFactory survives a rebind (_explicitFactory), so a
re-derived dialect cannot displace what a test passed in.
- FOCAS: options and backend move as a pair in InitializeAsync.
Re-resolving Sql's connection string on reinit is a side benefit: a rotated
credential is picked up without a process restart.
Drivers constructed directly get no rebinder and keep their constructor
binding, so every existing "{}"-passing lifecycle test is unaffected by
construction rather than by luck.
The tests pin ATOMICITY, not merely that a re-parse happened — a test checking
only options would have passed against the broken version. Confirmed by
simulating the half-fix (adopt options, skip the backend): 2 of 3 FOCAS tests
go red. Reverting Sql's rebind turns its connection-string test red. Sql also
asserts that a reinit which cannot resolve its new connection leaves the
previous binding whole rather than half-adopting.
All 12 drivers now honour a changed config in place. Sql.Tests 226, FOCAS.Tests
275 pass.
Bookkeeping: 13 .tasks.json files closed — the 8 the audit named, plus
mesh-phase4 (Task 9 landed 3a590a0c), the 4th stillpending phase file, and
historian-tcp-transport closed as OBSOLETE (it targets the retired Wonderware
sidecar; there is no Wonderware backend in the tree). Each carries a
closureNote naming the evidence rather than just a status flip.
11 archreview live gates written back from STATUS.md, which had recorded them
passed since 2026-07-13/15 without ever updating the task files: R2-01 #11,
R2-02 #15/#18, R2-05 T15, R2-06 T12, R2-07 T5/T6/T12/T14, R2-11 T22/T24.
R2-03 and R2-10 were deliberately NOT closed. STATUS.md's headline says "every
Round-2 live gate is now GREEN", but its own per-item detail says R2-03 is
"live-blocked on this data-less rig" and R2-10 verified the pipeline with
"breaker-OPEN state not forced". Both now carry an openNote recording the
actual blocker.
The point of the sweep: ~140 phantom pending tasks are gone, so the genuinely
unexecuted 19-task AdminUI follow-ups plan is now the largest open item in
docs/plans/ instead of being buried under work that shipped months ago.
Tier truth: docs/v2/driver-stability.md now states that its tier table is
aspirational — every driver runs Tier A because no factory passes a tier, so
MemoryRecycle and ScheduledRecycleScheduler have never engaged, and the config
parser validates a RecycleIntervalSeconds that does nothing. It also corrects a
staleness the audit missed: the separate-Windows-service hosting it describes
for Tier C no longer exists (Galaxy moved to the mxaccessgw sidecar in PR 7.2,
FOCAS went in-process with its managed wire client). docs/drivers/README.md
already said Tier A for both, so the two no longer contradict.
No runtime change — enabling recycle on two live drivers wants its own live
gate, not a docs-pass side effect. Keep-or-delete tracked as Gitea #522.
A parity guard already existed for DriverTypeNames <-> the /raw driver picker,
but not for the two dispatch maps downstream, so a driver could be registered,
offered in the picker, and then have no config form. That is exactly what
happened to Calculation (G-1) and to Sql/MTConnect/Calculation on the device
modal (G-2) — both survived review because nothing could see them.
The maps had to become data first. A Razor @switch compiles into
BuildRenderTree's IL, so no test can enumerate its cases; the picker guard
works only because RawDriverTypeDialog keeps its data in a field the markup
enumerates. Both modals now render from DriverConfigFormMap / DeviceFormMap
through <DynamicComponent>, and being public those maps need none of the
picker test's BindingFlags.NonPublic fragility.
- G-1 CalculationDriverForm.razor — RunTimeout was unauthorable via the UI.
- G-2 Sql/MTConnect/Calculation declared single-connection rather than given
hollow device forms; each holds one connection at the driver level.
- G-3 DriverConfigModal's hardcoded "Galaxy or Mqtt" -> IsSingleConnection, so
Sql/MTConnect authors are no longer sent to a device form with no endpoint.
- G-4 DriverFormMapParityTests (5, both directions) + DriverDispatchMapParityTests.
- G-5 CsvColumnMap entries for Sql, Mqtt, MTConnect.
- G-6 RawBrowseCommitMapper Sql branch — and the browser end, which the audit
missed: SqlBrowseSession emitted NO AddressFields, so a browsed leaf carried
only a column name, and a column alone cannot address a Sql tag. It now
travels schema/table/column and the mapper builds a WideRow config from them.
A branch alone would have produced a half-built blob.
Two findings while writing the guards. The G-6 test showed Modbus and
Calculation also hit the generic address fallback — correctly, as neither is
browsable — so it asserts the fall-through set EQUALS a documented
non-browsable list in both directions; a one-way check would let a newly
browsable driver be quietly added to the exclusion list. And
TagConfigDriverTypeNameGuardTests was hollow: it enumerated a hand-written
TheoryData that had already drifted (omitting Galaxy, Sql and Mqtt), guarding
renames but not gaps. Now enumerated from the map with a coverage test facing
the other way.
Live-verified on docker-dev, since this repo has no bUnit and no unit test
reaches Blazor parameter binding: opened Calculation's config (previously "No
typed config form"), typed 3500, saved, reopened — the value persisted, so the
DynamicComponent two-way binding round-trips. Sql's form renders fully and now
reads "This driver holds a single connection, authored above".
AdminUI.Tests 930 passed.
Five drivers ignored the config handed to them on reinitialize, serving the
options their constructor captured — and the deployment sealed green anyway.
Fixed on both halves, as chosen.
Seam (load-bearing): DriverSpawnPlanner routes a changed DriverConfig to
ToStop + ToSpawn instead of an in-place delta, making the factory the single
parse authority. This REVERSES the deliberate decision documented at
DriverSpawnPlan.cs:49-50 ("a pure DriverConfig change stays an in-place delta
— no reconnect"). That reasoning was correct about the resilience pipeline and
wrong about the driver. The accepted price is a reconnect per config edit.
Per-driver, and this split was not anticipated:
- Re-parse in place — Modbus, AbLegacy, OpcUaClient. ParseOptions extracted
from each factory, called from InitializeAsync behind a HasConfigBody guard
so "{}" still keeps the constructor options.
- Respawn-only, deliberately NOT re-parsed — Sql and FOCAS. Each builds more
than options from config (Sql's dialect + resolved connection string,
FOCAS's client-factory backend, both injected at construction), so adopting
new options alone would run a NEW tag set against an OLD connection. Half a
re-parse is worse than none. SqlDriver's doc-comment asserted the opposite
premise — "config parsing belongs to the factory, which builds a fresh
instance" — which was false when written and is true only now.
Two green seals removed from ApplyChildDelta: it overwrote the cached Spec
synchronously BEFORE the child dequeued the message, so the host believed the
new config was live and the next reconcile computed no delta, sealing the
drift permanently; and it Tell'd with no Receive<ApplyResult> registered, so a
failed reinit — including Galaxy's deliberate NotSupportedException —
dead-lettered.
Exposed (not created) by the change: a factory throw is a CONFIG error, and
SpawnChild catches it and silently substitutes a stub. Only a brand-new driver
could reach that before; an ordinary config edit can now. Raised Warning ->
Error with an actionable message. It still does not fail the deployment —
doing so would let one malformed driver block a fleet deploy, so that is a
follow-up rather than a drive-by.
Tests: every pre-existing reinit test in these suites passes "{}", the exact
input a guarded re-parser treats as "keep constructor options" — blind to this
defect by construction. New tests pass CHANGED json; the Modbus one was
verified falsifiable by deleting the re-parse (goes red). Two tests asserted
the old behaviour and were rewritten, including the positive control in
DriverHostActorUnreadableArtifactTests, whose observable moved from
UnsubscribeAsync to ShutdownAsync now that teardown happens by stopping the
child rather than emptying its desired set.
Remaining full-suite failures are pre-existing and environmental, verified
identical on the pre-change tree: Host.IntegrationTests (3) and
Driver.AbLegacy.IntegrationTests (4, docker fixture-gated).
#507 was filed as "injection inert in v3, re-migrate onto the raw subtree".
Reading it, the retained code is not revivable: it resolves equipment from
EquipmentNode.DriverInstanceId UNION EquipmentTags, and BOTH are structurally
empty in v3 (AddressSpaceComposer always constructs EquipmentNode with a null
DriverInstanceId; DeploymentArtifact hard-codes an empty EquipmentTags set).
Removing the guard would have changed a log line and injected nothing. So it
is deleted rather than fixed, and #507 closes as superseded by /raw
browse-commit.
Removed: HandleDiscoveredNodes, PartitionDiscoveredByDeviceHost,
ShouldWarnPartition, PlansRoutingEqual, ApplyDiscoveredPlansForDriver,
_discoveredByDriver, the redeploy re-inject tail, DiscoveredNodeMapper,
DiscoveredInjection, AddressSpaceApplier.MaterialiseDiscoveredNodes,
OpcUaPublishActor.MaterialiseDiscoveredNodes, and the Runtime-local copies of
CapturingAddressSpaceBuilder/DiscoveredNode.
The connect-time discovery loop goes too (StartDiscovery, RediscoverTick,
HandleRediscoverAsync, DiscoveredNodesReady, TriggerRediscovery). With
injection gone it had no consumer, and leaving it would either dead-letter or
keep browsing real devices up to ~15x per connect to drop the result.
ITagDiscovery itself stays — the /raw browse picker drives it through
Commons/Browsing/DiscoveryDriverBrowser, which never went through the actor,
so the picker is unaffected.
deferment.md §4.4 called the 18 DiscoveryInjectionDormantV3 tests "Real —
blocked on #507". They are not: they assert an equipment-rooted graft
(EquipmentRootNodeId == "EQ-1") that v3 cannot produce, so they could never
have unskipped as written. Deleted along with DiscoveredNodeMapperTests, the
CapturingAddressSpaceBuilder tests, and the MaterialiseDiscoveredNodes tests
in AddressSpaceApplierTests/OpcUaPublishActorTests. Runtime.Tests skipped:
31 -> 13.
IHostConnectivityProbe is deliberately NOT resolved here. GetHostStatuses()
has no production call site and nothing writes a DriverHostStatus row, but the
table was re-created in the v3 initial migration, so deleting on inference
would be wrong. Split to Gitea #521 with both options costed.
CLAUDE.md, docs/drivers/{Galaxy,TwinCAT,MTConnect}.md updated — they described
the seam as dead.
Build clean; Runtime.Tests 476 passed / 13 skipped; OpcUaServer.Tests 362
passed / 4 skipped.
deferment.md §8.1 asked for a decision: wire IPermissionEvaluator into the node
manager, or say plainly that ACLs are not enforced. Decision: the latter. The
evaluator stays in the tree; #520 tracks the wire-up.
Verifying the claim turned up four things worse than the audit recorded:
- docs/security.md, not ReadWriteOperations.md, was the highest-risk doc. It
claimed an anonymous session is "default-denie[d] any node a session has no
ACL grant for". The opposite holds: anonymous can Browse/Read/Subscribe/
HistoryRead the entire address space, and is refused writes and alarm acks
only because it carries no roles.
- Three of five documented data-plane role strings do nothing.
OpcUaDataPlaneRoles declares exactly WriteOperate and AlarmAck; ReadOnly,
WriteTune and WriteConfigure are compared against nowhere in src/. The doc
called all five "exact, case-insensitive, and code-true".
- ReadWriteOperations.md was fiction beyond the ACL claims. OnReadValue has
zero occurrences in src/ — the documented read path did not exist. Reads
never reach a driver: the node manager is push-model and the SDK serves a
Read from the cached pushed value. WriteAuthzPolicy, _sourceByFullRef,
_writeIdempotentByFullRef and IRoleBearer are likewise zero-hit, and
CapabilityInvoker is not referenced by the OpcUaServer project at all. This
file is rewritten rather than bannered.
- v2-release-readiness.md claimed a whole enforcement layer shipped —
FilterBrowseReferences, GateCallMethodRequests, MapCallOperation,
AuthorizationBootstrap, Node:Authorization:* keys. None exist. Whatever
landed on the v2 branch did not survive into the shipped tree.
The UI change is the operative one: ClusterAcls.razor and AclEdit.razor now
warn that a grant is saved and shipped in every deployment artifact but never
evaluated, so an operator cannot author a deny rule believing it works.
deferment.md gains a §9 execution log tracking §8 remediation.
Adds deferment.md — a source-verified inventory of new-driver status, all 17
open issues, in-code deferrals, open live gates and plan bookkeeping. Every
claim was checked against src/ and git, not against documentation.
Headline finding: three subsystems are authored, persisted and shipped but
never executed —
* Node ACLs: IPermissionEvaluator/TriePermissionEvaluator have zero
production consumers and OtOpcUaNodeManager never references them, yet
ClusterAcls.razor authors NodeAcl rows and ConfigComposer.cs:51 ships them
in every artifact. An authored deny rule has no effect.
* IRediscoverable / IHostConnectivityProbe raise into the void (#518/#507);
nothing ever writes a DriverHostStatus row.
* DriverTypeRegistry is vestigial, so no factory passes a tier and every
driver runs Tier A with the Tier-C protections dormant.
Also records the Calculation driver as picker-visible but unauthorable
(DriverConfigModal has no case) — the "registered but unauthorable" class
recurring after the Sql picker defect — and notes that no parity test guards
DriverConfigModal/DeviceModal, which is why it survived review.
Documentation corrections (source-verified):
* ReadWriteOperations.md claimed "a denied read never hits the driver" via
four types that do not exist in src/. Bannered + struck through; a security
review reading that page would have concluded a per-node ACL gate exists.
* CLAUDE.md: the Change Detection sentence was false (DriverHost consumes
nothing); the mesh Phase 4 and Phase 5 live gates and the auto-down 1-vs-1
gate had all PASSED; the ScriptedAlarmState table was already dropped.
* driver-expansion tracking: Modbus RTU and SQL poll are merged, not
pending — its command table would have made someone rebuild two shipped
drivers in a fresh worktree.
* drivers/README.md: dead DriverTypeRegistry paragraph, retired
SystemPlatform namespace kind, missing Sql + Calculation rows, missing
Modbus RTU-over-TCP transport.
* TwinCAT.md/Galaxy.md promised an address-space rebuild that never happens.
* Historian.md gained the #491 unproven-value-capture pointer.
* IncrementalSync.md and AddressSpace.md bannered as wholesale v2-era (the
rewrite is recorded as still owed, not done here); four v2 status docs
bannered as historical; three wrong-name-for-a-live-type fixes.
Left deliberately untouched: the secrets NoOpSecretReplicator line, which
makes a security claim and needs the real behaviour identified rather than a
rename.
Joseph Doherty