feat: v1.1 schema, 15 conformance fixtures, and corpus self-check

[sepehr-safari]

Seeds the repository with the three pieces promised in shiv3/ocpp-cp-simulator#188.

Schema

schema/trace-v1.schema.json is the v1.1 schema from the simulator's docs/trace-format.md, byte-identical apart from $id, which now points at this organization.

Fixtures

15 traces converted from the DebugKit scenario corpus (OCPP 1.6J, fully synthetic), each paired with an expected.json describing the consumer view a conformant implementation derives from it: correlation pairs, effective actions, unanswered calls, orphan responses.

Two properties worth knowing before review:

• Responses omit the optional action field on purpose, so reproducing expected.json requires actually implementing messageId correlation.
• raw is present on every record, and the self-check verifies it decodes to exactly the fields the record decomposes.

Coverage gaps are listed openly in fixtures/README.md: no CALLERROR records yet, no messageId reuse, single charge point, 1.6J only, no SOAP. @shiv3 the multi-CP and 2.0.1 traces you offered would close the biggest ones.

Conformance

conformance/README.md defines what conformance means for producers and consumers, including the correlation rule. conformance/validate.mjs is the corpus self-check and the reference consumer: schema validation, raw fidelity, and exact recomputation of every expected.json. npm test runs it, and CI runs it on every push and pull request, so a fixture and its expected view cannot drift apart unnoticed.

Notes

• License files are intentionally absent; they follow the decision in Define and apply a license for the Open OCPP Trace specification and assets #1, and everything here is contributed under whatever terms that lands on.
• The prose specification (the field semantics table) still lives in the simulator's docs/trace-format.md. Migrating it here is left to a follow-up so it lands under its author.
• The v1.1 open question about unparseable frames is filed as Representing frames that do not parse as OCPP-J #2, since it decides what a malformed-frame fixture would even look like.

[shiv3]

Thanks for turning this around so quickly — schema, corpus, and a runnable self-check in one go is exactly the seed this repo needed.

Verified the schema block is byte-identical to the simulator's docs/trace-format.md (only $id differs), and independently re-implemented the correlation algorithm (a different traversal than validate.mjs's backward scan) to cross-check all 15 expected.json files against their trace.jsonl — matches exactly, including the unresponsive-csms unanswered-call case. Solid corpus.

Two small things worth a look before/alongside the CALLERROR fixtures you flagged as a follow-up:

1. checkRawFidelity() in validate.mjs only checks error.code against the raw frame for CALLERROR, not error.description/error.details — unlike the full-payload check it does for CALL/CALLRESULT. Worth completing now so the first CALLERROR fixture can't slip a partial mismatch.
2. None of the 15 fixtures produce a non-empty orphanResponses, and that's not in fixtures/README.md's coverage-gap list (unlike CALLERROR, messageId reuse, etc.) — might be worth adding it there, or adding one small fixture that actually hits an orphan response, since it's one of the four consumer-view concepts the PR description leads with.

Minor/optional: the ok print in validate.mjs is gated on a global failures === 0, so on a red run later-but-passing fixtures stop printing their status line. Not a correctness issue, just a triage nit.

Approving — happy for the two items above to land here or as a follow-up, your call.

[sepehr-safari]

Thanks for the thorough pass, and especially for re-deriving the correlation with a different traversal and checking all 15 expected.json against it. Two independent implementations agreeing across the corpus is what this repo is for, so that cross-check is the part I care about most.

Both items are fair, and I'd rather close them here than defer:

1. I'll extend checkRawFidelity() to compare error.description and error.details against the raw CALLERROR frame, matching what it already does for CALL/CALLRESULT, so the first CALLERROR fixture can't slip a partial mismatch.
2. I'll add a small fixture that produces a real orphanResponses entry (a response with no preceding CALL), since it's one of the four consumer-view concepts and nothing currently exercises it, and keep the coverage-gap list honest.

I'll fold in the ok-print triage nit at the same time.

@juherr, this seeds the neutral repo so it's your call too. Happy to hold for your review, or to merge once those two fixtures are in if you're good with it.