FOCAS version-matrix stabilization (PR 1 of #220 split) — ship the cheap half of the hardware-free stability gap ahead of the Tier-C out-of-process split. Without any CNC or simulator on the bench, the highest-leverage move is to catch operator config errors at init time instead of at steady-state per-read. Adds FocasCncSeries enum (Unknown/16i/0i-D/0i-F family/30i family/PowerMotion-i) + FocasCapabilityMatrix static class that encodes the per-series documented ranges for macro variables (cnc_rdmacro/wrmacro), parameters (cnc_rdparam/wrparam), and PMC letters + byte ceilings (pmc_rdpmcrng/wrpmcrng) straight from the Fanuc FOCAS Developer Kit. FocasDeviceOptions gains a Series knob (defaults Unknown = permissive so pre-matrix configs don't break on upgrade). FocasDriver.InitializeAsync now calls FocasAddress.TryParse on every tag + runs FocasCapabilityMatrix.Validate against the owning device's declared series, throwing InvalidOperationException with a reason string that names both the series and the documented limit ("Parameter #30000 is outside the documented range [0, 29999] for Thirty_i") so an operator can tell whether the mismatch is in the config or in their declared CNC model. Unknown series skips validation entirely. Ships 46 new theory cases in FocasCapabilityMatrixTests.cs — covering every boundary in the matrix (widen 16i->0i-F: macro ceiling 999->9999, param 9999->14999; widen 0i-F->30i: PMC letters +K+T; PMC-number 16i=999/0i-D=1999/0i-F=9999/30i=59999), permissive Unknown-series behavior, rejection-message content, and case-insensitive PMC-letter matching. Widening a range without updating docs/v2/focas-version-matrix.md fails a test because every InlineData cites the row it reflects. Full FOCAS test suite stays at 165/165 passing (119 existing + 46 new). Also authors docs/v2/focas-version-matrix.md as the authoritative range reference with per-function citations, CNC-series era context, error-surface shape, and the link back to the matrix code; docs/v2/implementation/focas-isolation-plan.md as the multi-PR plan for #220 Tier-C isolation (Shared contracts -> Host skeleton -> move Fwlib32 calls -> Supervisor+respawn -> MMF+ops glue, 2200-3200 LOC across 5 PRs mirroring the Galaxy Tier-C topology); and promotes docs/drivers/FOCAS-Test-Fixture.md from "version-matrix coverage = no" to explicit coverage via the new test file + cross-links to the matrix and isolation-plan docs. Leaves task #220 open since isolation itself (the expensive half) is still ahead.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-04-20 13:44:37 -04:00
parent 64bbc12e8e
commit a6be2f77b5
8 changed files with 695 additions and 2 deletions
--- a/docs/v2/focas-version-matrix.md
+++ b/docs/v2/focas-version-matrix.md
@@ -0,0 +1,145 @@
+# FOCAS version / capability matrix
+
+Authoritative source for the per-CNC-series ranges that
+[`FocasCapabilityMatrix`](../../src/ZB.MOM.WW.OtOpcUa.Driver.FOCAS/FocasCapabilityMatrix.cs)
+enforces at driver init time. Every row cites the Fanuc FOCAS Developer
+Kit function whose documented input range determines the ceiling.
+
+**Why this exists** — we have no FOCAS hardware on the bench and no
+working simulator. Fwlib32 returns `EW_NUMBER` / `EW_PARAM` when you
+hand it an address outside the controller's supported range; the
+driver would map that to a per-read `BadOutOfRange` at steady state.
+Catching at `InitializeAsync` with this matrix surfaces operator
+typos + mismatched series declarations as config errors before any
+session is opened, which is the only feedback loop available without
+a live CNC to read against.
+
+**Who declares the series** — `FocasDeviceOptions.Series` in
+`appsettings.json`. Defaults to `Unknown`, which is permissive — every
+address passes validation. Pre-matrix configs don't break on upgrade.
+
+---
+
+## Series covered
+
+| Enum value | Controller family | Typical era |
+| --- | --- | --- |
+| `Unknown` | (legacy / not declared) | permissive fallback |
+| `Sixteen_i` | 16i / 18i / 21i | 1997-2008 |
+| `Zero_i_D` | 0i-D | 2008-2013 |
+| `Zero_i_F` | 0i-F | 2013-present, general-purpose |
+| `Zero_i_MF` | 0i-MF | 0i-F lathe variant |
+| `Zero_i_TF` | 0i-TF | 0i-F turning variant |
+| `Thirty_i` | 30i-A / 30i-B | 2007-present, high-end |
+| `ThirtyOne_i` | 31i-A / 31i-B | 30i simpler variant |
+| `ThirtyTwo_i` | 32i-A / 32i-B | 30i compact |
+| `PowerMotion_i` | Power Motion i-A / i-MODEL A | motion-only controller |
+
+## Macro variable range (`cnc_rdmacro` / `cnc_wrmacro`)
+
+Common macros `1-33` + `100-199` + `500-999` are universal across all
+series. Extended macros (`#10000+`) exist only on higher-end series.
+The numbers below reflect the extended ceiling per series per the
+DevKit range tables.
+
+| Series | Min | Max | Notes |
+| --- | ---: | ---: | --- |
+| `Sixteen_i` | 0 | 999 | legacy ceiling — no extended |
+| `Zero_i_D` | 0 | 999 | 0i-D still at legacy ceiling |
+| `Zero_i_F` / `Zero_i_MF` / `Zero_i_TF` | 0 | 9999 | extended added on 0i-F |
+| `Thirty_i` / `ThirtyOne_i` / `ThirtyTwo_i` | 0 | 99999 | full extended set |
+| `PowerMotion_i` | 0 | 999 | atypical — limited macro coverage |
+
+## Parameter range (`cnc_rdparam` / `cnc_wrparam`)
+
+| Series | Min | Max |
+| --- | ---: | ---: |
+| `Sixteen_i` | 0 | 9999 |
+| `Zero_i_D` / `Zero_i_F` / `Zero_i_MF` / `Zero_i_TF` | 0 | 14999 |
+| `Thirty_i` / `ThirtyOne_i` / `ThirtyTwo_i` | 0 | 29999 |
+| `PowerMotion_i` | 0 | 29999 |
+
+## PMC letters (`pmc_rdpmcrng` / `pmc_wrpmcrng`)
+
+Addresses are letter + number (e.g. `R100`, `F50.3`). Legacy
+controllers omit the `F`/`G` signal groups that 30i-family ladder
+programs use, and only the 30i-family exposes `K` (keep-relay) +
+`T` (timer).
+
+| Letter | 16i | 0i-D | 0i-F family | 30i family | Power Motion-i |
+| --- | :-: | :-: | :-: | :-: | :-: |
+| `X` | yes | yes | yes | yes | yes |
+| `Y` | yes | yes | yes | yes | yes |
+| `R` | yes | yes | yes | yes | yes |
+| `D` | yes | yes | yes | yes | yes |
+| `E` | — | yes | yes | yes | — |
+| `A` | — | yes | yes | yes | — |
+| `F` | — | — | yes | yes | — |
+| `G` | — | — | yes | yes | — |
+| `M` | — | — | yes | yes | — |
+| `C` | — | — | yes | yes | — |
+| `K` | — | — | — | yes | — |
+| `T` | — | — | — | yes | — |
+
+Letter match is case-insensitive. `FocasAddress.PmcLetter` is carried
+as a string (not char) so the matrix can do ordinal-ignore-case
+comparison.
+
+## PMC address-number ceiling
+
+PMC addresses are byte-addressed on read + bit-addressed on write;
+`FocasAddress` carries the bit index separately, so these are byte
+ceilings.
+
+| Series | Max byte | Notes |
+| --- | ---: | --- |
+| `Sixteen_i` | 999 | legacy |
+| `Zero_i_D` | 1999 | doubled since 16i |
+| `Zero_i_F` family | 9999 | |
+| `Thirty_i` family | 59999 | highest density |
+| `PowerMotion_i` | 1999 | |
+
+## Error surface
+
+When a tag fails validation, `FocasDriver.InitializeAsync` throws
+`InvalidOperationException` with a message of the form:
+
+```
+FOCAS tag '<name>' (<address>) rejected by capability matrix: <reason>
+```
+
+`<reason>` is the verbatim string from `FocasCapabilityMatrix.Validate`
+and always names the series + the documented limit so the operator
+can either raise the limit (if wrong) or correct the CNC series they
+declared (if mismatched). Sample:
+
+```
+FOCAS tag 'X_axis_macro_ext' (MACRO:50000) rejected by capability
+matrix: Macro variable #50000 is outside the documented range
+[0, 9999] for Zero_i_F.
+```
+
+## How this matrix stays honest
+
+- Every row is covered by a parameterized test in
+  [`FocasCapabilityMatrixTests.cs`](../../tests/ZB.MOM.WW.OtOpcUa.Driver.FOCAS.Tests/FocasCapabilityMatrixTests.cs)
+  — 46 cases across macro / parameter / PMC-letter / PMC-number
+  boundaries + unknown-series permissiveness + rejection-message
+  content + case-insensitivity.
+- Widening or narrowing a range in the matrix without updating this
+  doc will fail a test, because the theories cite the specific row
+  they reflect in their `InlineData`.
+- The matrix is not comprehensive — it encodes only the subset of
+  FOCAS surface the driver currently exposes (Macro / Parameter /
+  PMC). When the driver gains a new capability (e.g. tool management,
+  alarm history), add its series-specific range tables here + matching
+  tests at the same time.
+
+## Follow-up
+
+This validation closes the cheap half of the FOCAS hardware-free
+stability gap — config errors now fail at load instead of per-read.
+The expensive half is Tier-C process isolation so that a crashing
+`Fwlib32.dll` doesn't take the main OPC UA server down with it. See
+[`docs/v2/implementation/focas-isolation-plan.md`](implementation/focas-isolation-plan.md)
+for that plan (task #220).
--- a/docs/v2/implementation/focas-isolation-plan.md
+++ b/docs/v2/implementation/focas-isolation-plan.md
@@ -0,0 +1,163 @@
+# FOCAS Tier-C isolation — plan for task #220
+
+> **Status**: DRAFT — not yet started. Tracks the multi-PR work to
+> move `Fwlib32.dll` behind an out-of-process host, mirroring the
+> Galaxy Tier-C split in [`phase-2-galaxy-out-of-process.md`](phase-2-galaxy-out-of-process.md).
+>
+> **Pre-reqs shipped** (this PR): version matrix + pre-flight
+> validation + unit tests. Those close the cheap half of the
+> hardware-free stability gap. Tier-C closes the expensive half.
+
+## Why isolate
+
+`Fwlib32.dll` is a proprietary Fanuc library with no source, no
+symbols, and a documented habit of crashing the hosting process on
+network errors, malformed responses, and during handle recycling.
+Today the FOCAS driver runs in-process with the OPC UA server —
+a crash inside the Fanuc DLL takes every driver down with it,
+including ones that have nothing to do with FOCAS. Galaxy has the
+same class of problem and solved it with the Tier-C pattern (host
+service + proxy driver + named-pipe IPC); FOCAS should follow that
+playbook.
+
+## Topology (target)
+
+```
+-------------------------------------+         +--------------------------+
+|  OtOpcUa.Server (.NET 10 x64)       |         | OtOpcUaFocasHost         |
+|                                     |  pipe   | (.NET 4.8 x86 Windows    |
+|  ZB.MOM.WW.OtOpcUa.Driver.FOCAS     | <-----> |  service)                |
+|    - FocasProxyDriver (in-proc)     |         |                          |
+|    - supervisor / respawn / BackPr  |         |  Fwlib32.dll + session   |
+|                                     |         |  handles + STA thread    |
+-------------------------------------+         +--------------------------+
+```
+
+Why .NET 4.8 x86 for the host: `Fwlib32.dll` ships as 32-bit only.
+The Galaxy.Host is already .NET 4.8 x86 for the same reason
+(MXAccess COM bitness), so the NSSM wrapper pattern transfers
+directly.
+
+## Three new projects
+
+| Project | TFM | Role |
+| --- | --- | --- |
+| `ZB.MOM.WW.OtOpcUa.Driver.FOCAS.Shared` | `netstandard2.0` | MessagePack DTOs — `FocasReadRequest`, `FocasReadResponse`, `FocasSubscribeRequest`, `FocasPmcBitWriteRequest`, etc. Same assembly referenced by .NET 10 + .NET 4.8 so the wire format stays identical. |
+| `ZB.MOM.WW.OtOpcUa.Driver.FOCAS.Host` | `net48` x86 | Windows service. Owns the Fwlib32 session handles + STA thread + handle-recycling loop. Pipe server + per-call auth (same ACL + caller SID + shared secret pattern as Galaxy.Host). |
+| `ZB.MOM.WW.OtOpcUa.Driver.FOCAS` (existing) | `net10.0` | Collapses to a proxy that forwards each `IReadable` / `IWritable` / `ISubscribable` call over the pipe. `FocasCapabilityMatrix` + `FocasAddress` stay here — pre-flight runs before any IPC. |
+
+## Supervisor responsibilities (in the Proxy)
+
+Mirrors Galaxy.Proxy 1:1:
+
+1. Start the Host process on first `InitializeAsync` (NSSM-wrapped
+   service in production, direct spawn in dev) + heartbeat every
+   5s.
+2. If heartbeat misses 3× in a row, fan out `BadCommunicationError`
+   to every subscription and respawn with exponential backoff
+   (1s / 2s / 4s / max 30s).
+3. Crash-loop circuit breaker: 5 respawns in 60s → drop to
+   `BadDeviceFailure` steady state until operator resets.
+4. Post-mortem MMF: on Host exit, Host writes its last-N operations
+   + session state to an MMF the Proxy reads to log context.
+
+## IPC surface (approximate)
+
+Every `FocasDriver` method that today calls into Fwlib32 directly
+becomes an `ExecuteAsync` call with a typed request:
+
+| Today (in-process) | Tier-C (IPC) |
+| --- | --- |
+| `FocasTagReader.Read(tag)` | `client.Execute(new FocasReadRequest(session, address))` |
+| `FocasTagWriter.Write(tag, value)` | `client.Execute(new FocasWriteRequest(...))` |
+| `FocasPmcBitRmw.Write(tag, bit, value)` | `client.Execute(new FocasPmcBitWriteRequest(...))` — RMW happens in Host so the critical section stays on one process |
+| `FocasConnectivityProbe.ProbeAsync` | `client.Execute(new FocasProbeRequest())` |
+| `FocasSubscriber.Subscribe(tags)` | `client.Execute(new FocasSubscribeRequest(tags))` — Host owns the poll loop + streams changes back as `FocasDataChangedNotification` over the pipe |
+
+Subscription streaming is the non-obvious piece: the Host polls on
+its own timer + pushes change notifications so the Proxy doesn't
+round-trip per poll. Matches `Driver.Galaxy.Host` subscription
+forwarding.
+
+## PR sequence (proposed)
+
+1. **PR A — shared contracts**
+   Create `Driver.FOCAS.Shared` with the MessagePack DTOs. No
+   behaviour change. ~200 LOC + round-trip tests for each DTO.
+2. **PR B — Host project skeleton**
+   Create `Driver.FOCAS.Host` .NET 4.8 x86 project, NSSM wrapper,
+   pipe server scaffold with the same ACL + caller-SID + shared
+   secret plumbing as Galaxy.Host. No Fwlib32 wiring yet — returns
+   `NotImplemented` for everything. ~400 LOC.
+3. **PR C — Move Fwlib32 calls into Host**
+   Move `FocasNativeSession`, `FocasTagReader`, `FocasTagWriter`,
+   `FocasPmcBitRmw` + the STA thread into the Host. Proxy forwards
+   over IPC. This is the biggest PR — probably 800-1500 LOC of
+   move-with-translation. Existing unit tests keep passing because
+   `IFocasTagFactory` is the DI seam the tests inject against.
+4. **PR D — Supervisor + respawn**
+   Proxy-side heartbeat + respawn + crash-loop circuit breaker +
+   BackPressure fan-out on Host death. ~500 LOC + chaos tests.
+5. **PR E — Post-mortem MMF + operational glue**
+   MMF writer in Host, reader in Proxy. Install scripts for the
+   new `OtOpcUaFocasHost` Windows service. Docs. ~300 LOC.
+
+Total estimate: 2200-3200 LOC across 5 PRs. Consistent with Galaxy
+Tier-C but narrower since FOCAS has no Historian + no alarm
+history.
+
+## Testing without hardware
+
+Same constraint as today: no CNC, no simulator. The isolation work
+itself is verifiable without Fwlib32 actually being called:
+
+- **Pipe contract**: PR A's MessagePack round-trip tests cover every
+  DTO.
+- **Supervisor**: PR D uses a `FakeFocasHost` stub that can be told
+  to crash, hang, or miss heartbeats. The supervisor's respawn +
+  circuit-breaker behaviour is fully testable against the stub.
+- **IPC ACL + auth**: reuse the Galaxy.Host's existing test harness
+  pattern — negative tests attempt to connect as the wrong user and
+  assert rejection.
+- **Fwlib32 integration itself**: still untestable without hardware.
+  When a real CNC becomes available, the smoke tests already
+  scaffolded in `tests/ZB.MOM.WW.OtOpcUa.Driver.FOCAS.IntegrationTests/`
+  run against it via `FOCAS_ENDPOINT`.
+
+## Decisions to confirm before starting
+
+- **Sharing transport code with Galaxy.Host** — should the pipe
+  server + ACL + shared-secret + MMF plumbing go into a common
+  `Core.Hosting.Tier-C` project both hosts reference? Probably yes;
+  deferred until PR B is drafted because the right abstraction only
+  becomes visible after two uses.
+- **Handle-recycling cadence** — Fwlib32 session handles leak
+  memory over weeks per the Fanuc-published defect list. Galaxy
+  recycles MXAccess handles on a 24h timer; FOCAS should mirror but
+  the trigger point (idle vs scheduled) needs operator input.
+- **Per-CNC Host process vs one Host serving N CNCs** — one-per-CNC
+  isolates blast radius but scales poorly past ~20 machines; shared
+  Host scales but one bad CNC can wedge the lot. Start with shared
+  Host + document the blast-radius trade; revisit if operators hit
+  it.
+
+## Non-goals
+
+- Simulator work. `open_focas` + other OSS FOCAS simulators are
+  untested + not maintained; not worth chasing vs. waiting for real
+  hardware.
+- Changing the public `FocasDriverOptions` shape beyond what
+  already shipped (the `Series` knob). Operator config continues to
+  look the same after the split — the Tier-C topology is invisible
+  from `appsettings.json`.
+- Historian / long-term history integration. FOCAS driver doesn't
+  implement `IHistoryProvider` + there's no plan to add it.
+
+## References
+
+- [`docs/v2/implementation/phase-2-galaxy-out-of-process.md`](phase-2-galaxy-out-of-process.md)
+  — the working Tier-C template this plan follows.
+- [`docs/drivers/FOCAS-Test-Fixture.md`](../../drivers/FOCAS-Test-Fixture.md)
+  — what's covered today + what stays blocked on hardware.
+- [`docs/v2/focas-version-matrix.md`](../focas-version-matrix.md) —
+  the capability matrix that pre-flights configs before IPC runs.