docs(drivers): un-stale the MQTT README rows + track every known gap as an issue
The cross-driver README still described MQTT as pre-P2 — "Sparkplug B is P2
(`mode: SparkplugB` is a placeholder)" and "Sparkplug B has no fixture yet (P2)".
Both were false as of the P2 merge (c3a2b0f7): SparkplugB is a fully implemented
mode and the live suite is 15 tests (7 plain + 8 Sparkplug) against a
project-owned C# edge-node simulator. Mqtt.md and the scadaproj index were both
already correct; only this file missed the P2 sweep.
Filed the nine Known-gaps rows as tracking issues and linked them, so the table
stops being the only record of them:
#507 IRediscoverable is inert in v3 (platform gap — also Galaxy/TwinCAT/AbCip)
#508 write-through (IWritable) — NCMD/DCMD + plain publish
#509 a metric name containing '/' cannot be browse-committed
#510 _canRebirth captured at browse-open — reaped session still draws button
#511 Sparkplug primary-host STATE publishing (ActAsPrimaryHost)
#512 .Browser references .Driver — extract the client-options builder
#513 editor writes a meaningless payloadFormat into a Sparkplug blob
#514 Request-rebirth unarmable on a plant that birthed before the window
#515 DataSet / Template / PropertySet metric support
#507 is the one worth reading: it is a v3 platform gap, not an MQTT bug —
DriverHostActor.HandleDiscoveredNodes hard-returns, so every IRediscoverable
driver fires into a void and the edge gaining a point needs a redeploy to show up.
No code change; docs + issue links only.
This commit is contained in:
Joseph Doherty
+19
-14
@@ -14,8 +14,9 @@ for the design.
|
||||
|
||||
> **Status.** P1 (plain MQTT) and P2 (Sparkplug B ingest) are both shipped and live-gated against the
|
||||
> Mosquitto + Sparkplug-simulator fixture. Write-through (`IWritable` — NCMD/DCMD/plain publish) is
|
||||
> **deferred**; every MQTT node materializes read-only. See
|
||||
> [Known gaps](#known-gaps) for what is wired but not yet end-to-end observable.
|
||||
> **deferred** ([#508](https://gitea.dohertylan.com/dohertj2/lmxopcua/issues/508)); every MQTT node
|
||||
> materializes read-only. See [Known gaps](#known-gaps) — every entry there carries a tracking issue
|
||||
> (#507–#515).
|
||||
|
||||
## Project Layout
|
||||
|
||||
@@ -28,8 +29,9 @@ for the design.
|
||||
> `.Browser` references `.Driver` (not just `.Contracts`) — a **documented, deliberate exception**
|
||||
> so browse can reuse `MqttConnection.BuildClientOptions` (TLS / CA-pin / credentials / protocol
|
||||
> mapping). Its cost is pulling `Core` into the AdminUI graph. The csproj carries a boxed warning;
|
||||
> do not copy the pattern into a new driver's browser. The clean fix (deferred) is a leaf
|
||||
> `Driver.Mqtt.Transport` project.
|
||||
> do not copy the pattern into a new driver's browser. The clean fix — extract the pure
|
||||
> options builder down into `.Contracts` — is tracked as
|
||||
> [#512](https://gitea.dohertylan.com/dohertj2/lmxopcua/issues/512).
|
||||
|
||||
## Capability Surface
|
||||
|
||||
@@ -380,13 +382,16 @@ identical words. See [TestConnectProbes.md](TestConnectProbes.md).
|
||||
|
||||
## Known gaps
|
||||
|
||||
| Gap | Effect | Notes |
|
||||
|---|---|---|
|
||||
| **Write-through (`IWritable`)** | Every MQTT node is read-only | Deferred to P3 — NCMD/DCMD + plain publish, optimistic-Good + self-correct on echo |
|
||||
| **Rediscovery is inert** | A DBIRTH introducing a metric does not change the OPC UA tree; redeploy to pick it up | v3 platform gap — nothing subscribes to `OnRediscoveryNeeded` and `DriverHostActor.HandleDiscoveredNodes` hard-returns. Driver-side decision is log-observable |
|
||||
| **Rebirth needs a non-empty tree** | `Request rebirth…` cannot be armed on a plant that has not birthed since the window opened | See [Refresh](#refresh). Session side already supports a bare `{group}/{edgeNode}` scope |
|
||||
| **`_canRebirth` is captured at browse-open** | A session reaped for idleness (2 min) still draws the button, and using it errors | The error is shown, not swallowed; reopen the browser |
|
||||
| **Primary-host STATE publishing** | `ActAsPrimaryHost` does nothing | Logs a startup warning rather than being silently inert — see the `Sparkplug` sub-object table |
|
||||
| **A metric whose name contains `/` cannot be browse-committed** | The commit is refused **whole**, in words (`Row N: Name must not contain '/'`) | The derived **tag Name** is a RawPath segment and may not contain `/`, while `metricName` legitimately may. This blocks the spec-mandatory `Node Control/Rebirth`. **Workaround: Manual entry** — author the tag with a `/`-free name and type the slashed `metricName` into the Sparkplug editor, which accepts it. Nothing is silently mis-bound |
|
||||
| **`DataSet` / `Template` / `PropertySet` metrics** | Skipped with a warning | v1-unsupported; never coerced into a guessed type |
|
||||
| **The tag editor writes `payloadFormat` into a Sparkplug blob** | Cosmetic only | `payloadFormat` is Plain-only; the factory routes on the Sparkplug keys and ignores it. Noise in the blob, not a mis-binding |
|
||||
Every row is tracked. None is a silent failure — each either refuses in words or logs.
|
||||
|
||||
| Gap | Issue | Effect | Notes |
|
||||
|---|:---:|---|---|
|
||||
| **Write-through (`IWritable`)** | [#508](https://gitea.dohertylan.com/dohertj2/lmxopcua/issues/508) | Every MQTT node is read-only | Deferred to P3 — NCMD/DCMD + plain publish, optimistic-Good + self-correct on echo |
|
||||
| **Rediscovery is inert** | [#507](https://gitea.dohertylan.com/dohertj2/lmxopcua/issues/507) | A DBIRTH introducing a metric does not change the OPC UA tree; redeploy to pick it up | v3 **platform** gap, not MQTT-specific — nothing subscribes to `OnRediscoveryNeeded` and `DriverHostActor.HandleDiscoveredNodes` hard-returns. Also affects Galaxy / TwinCAT / AbCip. Driver-side decision is log-observable |
|
||||
| **Rebirth needs a non-empty tree** | [#514](https://gitea.dohertylan.com/dohertj2/lmxopcua/issues/514) | `Request rebirth…` cannot be armed on a plant that has not birthed since the window opened | See [Refresh](#refresh). Session side already supports a bare `{group}/{edgeNode}` scope |
|
||||
| **`_canRebirth` is captured at browse-open** | [#510](https://gitea.dohertylan.com/dohertj2/lmxopcua/issues/510) | A session reaped for idleness (2 min) still draws the button, and using it errors | The error is shown, not swallowed; reopen the browser |
|
||||
| **Primary-host STATE publishing** | [#511](https://gitea.dohertylan.com/dohertj2/lmxopcua/issues/511) | `ActAsPrimaryHost` does nothing | Logs a startup warning rather than being silently inert — see the `Sparkplug` sub-object table |
|
||||
| **A metric whose name contains `/` cannot be browse-committed** | [#509](https://gitea.dohertylan.com/dohertj2/lmxopcua/issues/509) | The commit is refused **whole**, in words (`Row N: Name must not contain '/'`) | The derived **tag Name** is a RawPath segment and may not contain `/`, while `metricName` legitimately may. This blocks the spec-mandatory `Node Control/Rebirth`. **Workaround: Manual entry** — author the tag with a `/`-free name and type the slashed `metricName` into the Sparkplug editor, which accepts it. Nothing is silently mis-bound |
|
||||
| **`DataSet` / `Template` / `PropertySet` metrics** | [#515](https://gitea.dohertylan.com/dohertj2/lmxopcua/issues/515) | Skipped with a warning | v1-unsupported; never coerced into a guessed type |
|
||||
| **The tag editor writes `payloadFormat` into a Sparkplug blob** | [#513](https://gitea.dohertylan.com/dohertj2/lmxopcua/issues/513) | Cosmetic only | `payloadFormat` is Plain-only; the factory routes on the Sparkplug keys and ignores it. Noise in the blob, not a mis-binding |
|
||||
| **`.Browser` references `.Driver`, not just `.Contracts`** | [#512](https://gitea.dohertylan.com/dohertj2/lmxopcua/issues/512) | A documented layering exception | Taken to reuse `MqttConnection.BuildClientOptions` rather than duplicate TLS/CA-pin logic — see the note at the top of this doc |
|
||||
|
||||
Reference in New Issue
Block a user