test(conformance): vendored scenarios + golden harness, close out S2

AGENTS.md

@@ -20,7 +20,8 @@ studio/
 ├── src/
 │   ├── main.zig             # Model, Msg, update, app wiring
 │   ├── app.native           # the view (declarative markup)
-│   ├── ocpp/                # pure, headless OCPP engine (types, parser, …)
+│   ├── ocpp/                # pure, headless OCPP engine (types, parser, detection, …)
+│   │   └── conformance/     # vendored shared-contract fixtures + goldens + harness
 │   └── tests.zig            # headless view/model tests
 ├── scripts/smoke.sh         # portable automation smoke test (Xvfb in CI)
 ├── docs/adr/                # architecture decision records

CURRENT_STATE.md

@@ -8,8 +8,8 @@
 
 ## Active milestone
 
-**S2 — Detection + conformance (0.1.0): in progress.** S1 — Engine core is
-complete (see [ROADMAP.md](ROADMAP.md)).
+**S2 — Detection + conformance (0.1.0): complete.** Next up: **S3 — Inspector UI**
+(see [ROADMAP.md](ROADMAP.md)).
 
 ## What's done
 
@@ -53,22 +53,26 @@ conformance contract (behavior, not source) and tested via
 correlates end to end (one completed session, transactionId 100001); engine
 tests green headlessly on macOS + Linux.
 
+### S2 — Detection + conformance ✅
+
+The full 16-rule OCPP 1.6J failure taxonomy in `src/ocpp/detection.zig`, mirroring
+the toolkit's `detection.ts`, plus a harness that locks Studio's output to the
+toolkit's:
+
+- **Failure model + foundational rules** (#19) — the `Failure` / `FailureCode` /
+  `FailureSeverity` model and rules 1–3.
+- **Protocol & transaction rules** (#20) — rules 4–10.
+- **Timing & anomaly rules** (#21) — rules 11–16.
+- **Conformance harness** (#22) — 15 vendored scenario traces + goldens
+  (`src/ocpp/conformance/`, `contract-v1`, generated from the toolkit) and a
+  `native test` gate asserting Studio's de-duplicated, sorted `FailureCode` set
+  matches each golden.
+
+**Exit criteria met:** 15/15 scenarios match the locked goldens.
+
 ## What's in progress
 
-**S2 — Detection + conformance (0.1.0).** Building the 16-rule failure taxonomy
-in Zig, mirroring the toolkit's `detection.ts`:
-
-- **Failure model + foundational rules** (#19) — the `Failure` / `FailureCode`
-  (16-code taxonomy) / `FailureSeverity` model, the `detection.zig` orchestrator
-  with per-code severity + remediation metadata, and rules 1–3
-  (FAILED_AUTHORIZATION, CONNECTOR_FAULT, STATION_OFFLINE_DURING_SESSION).
-- **Protocol & transaction rules** (#20) — rules 4–10: TIMEOUT_NO_HEARTBEAT,
-  METER_VALUE_GAP, INVALID_STOP_REASON, UNEXPECTED_START,
-  STATUS_TRANSITION_VIOLATION, DIAGNOSTICS_FAILURE, FIRMWARE_UPDATE_FAILURE.
-- **Timing & anomaly rules** (#21) — rules 11–16: SUSPICIOUS_SESSION_DURATION,
-  SLOW_RESPONSE, HEARTBEAT_INTERVAL_VIOLATION, METER_VALUE_ANOMALY,
-  UNRESPONSIVE_CSMS, REPEATED_BOOT_NOTIFICATION. All 16 rules now run.
-- Next: the conformance harness over the 15 shared scenarios (#22).
+- Nothing in flight — S2 is closed; S3 is next.
 
 ## What's next
 
@@ -86,8 +90,8 @@ search / filter, handling traces far larger than a browser tab can hold.
 | --- | --- |
 | `repo` (tooling, CI) | ✅ done for S0 |
 | `docs` (docs, ADRs) | ✅ done for S0 |
-| `ocpp` (engine) | 🚧 detection in progress (S2) |
+| `ocpp` (engine) | ✅ done for S2 |
 | `ui` (native views) | ⬜ placeholder (S3) |
 | `capture` (live proxy) | ⬜ not started (S5) |
 | `cli` (headless) | ⬜ not started (S4) |
-| `conformance` | ⬜ not started (S2) |
+| `conformance` | ✅ done for S2 (15/15, `contract-v1`) |

ROADMAP.md

@@ -10,8 +10,8 @@ the stable part.
 | --- | --- | --- | --- |
 | **S0** | 0.0.x | Foundation | ✅ Complete |
 | **S1** | 0.1.0 | Engine core | ✅ Complete |
-| **S2** | 0.1.0 | Detection + conformance | Next up |
-| **S3** | 0.2.0 | Inspector UI | Planned |
+| **S2** | 0.1.0 | Detection + conformance | ✅ Complete |
+| **S3** | 0.2.0 | Inspector UI | Next up |
 | **S4** | 0.3.0 | Analysis parity+ | Planned |
 | **S5** | 0.4.0 | Live capture ⭐ | Planned |
 | **S6** | 1.0.0 | 1.0 polish | Planned |
@@ -32,11 +32,13 @@ inference), and session correlation by transaction id. **Exit criteria met:** th
 vendored `normal-session` fixture parses and correlates end to end; engine tests
 run green headlessly (`native test -Dplatform=null`) on macOS + Linux.
 
-## S2 — Detection + conformance (0.1.0)
+## S2 — Detection + conformance (0.1.0) ✅
 
-The full OCPP 1.6J failure taxonomy in Zig, plus the conformance harness that
-runs the shared scenario fixtures and checks Studio's detected failures against
-locked goldens. **Exit:** the shared scenarios match, byte for byte.
+The full OCPP 1.6J failure taxonomy in Zig — all 16 detection rules — plus the
+conformance harness that runs the 15 shared scenario fixtures and checks Studio's
+detected failures against locked goldens generated from the toolkit.
+**Exit criteria met:** 15/15 scenarios match the `contract-v1` goldens under
+`native test`.
 
 ## S3 — Inspector UI (0.2.0)
 

src/ocpp/conformance/README.md

@@ -0,0 +1,44 @@
+# Conformance contract (`contract-v1`)
+
+Studio and the TypeScript toolkit are **independent implementations** of the same
+OCPP analysis behavior (ADR-0001). This directory is the machine-checked contract
+between them: the shared scenarios and the failure codes each must detect.
+
+- **`fixtures/<name>.json`** — the 15 shared scenario **traces** (the input).
+- **`goldens/<name>.json`** — each scenario's expected **de-duplicated, sorted
+  `FailureCode` set** (the output to match), as a JSON array of wire codes.
+- **`harness.zig`** — runs, under `native test`, the full engine
+  (`parseTrace → buildSessionTimeline → detectFailures`) over every fixture and
+  asserts its detected code set equals the golden — the same comparison the
+  toolkit's `evaluateScenario` makes. A missing or extra code fails the scenario
+  by name.
+
+## Generated, not hand-authored
+
+Both the fixtures and the goldens are exported from the toolkit — the source of
+truth — so a golden can never drift from what the reference implementation
+actually detects. They are regenerated (not edited by hand) from the toolkit's
+`scenarios` and `evaluateScenario`:
+
+```js
+// run against a built @ocpp-debugkit/toolkit
+import { scenarios } from '@ocpp-debugkit/toolkit/scenarios';
+import { evaluateScenario } from '@ocpp-debugkit/toolkit';
+for (const s of scenarios) {
+  const detected = [...new Set(evaluateScenario(s).failures.map((f) => f.code))].sort();
+  // fixtures/<s.name>.json  <- JSON.stringify(s.trace)
+  // goldens/<s.name>.json   <- JSON.stringify(detected)
+}
+```
+
+## Why it lives under `src/`
+
+The Native SDK zero-config build (ADR-0004) embeds test data with `@embedFile`,
+which resolves within the source module root. Keeping the vendored contract here
+lets the harness embed it directly — no runtime file I/O, fully hermetic.
+
+## Versioning
+
+Tagged **`contract-v1`** (OCPP 1.6J). When the contract changes — new rules,
+new scenarios, or OCPP 2.0.1 — regenerate against the matching toolkit release
+and bump the tag.