info-tech-canon/docs/emission-cadence-adoption.md
tegwick 038677ab53 Record net-kingdom's Emission Cadence declaration and fix the stale digest (INFO-WP-0029)
net-kingdom published the first source-owned declaration (116643f). It passes
emission-review. Record it as feedback, and file its session-scoped-silence
incompatibility as demand/EmissionActivityScope.md. T02 is done, and T04 is in
progress until activity-core answers.

The adoption brief sent owners the 0.1.0 draft digest, because the candidate
promotion changed the standard's text after the digest was taken. The export
manifest also hard-coded status "draft". The manifest now derives status and
version from the standard, a test pins that, and the brief names the candidate
digest b08b4d95fc4b0bd3. The wire schema is unchanged.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 121421@bnt-lap001
Assistant-Session: 36f8657c-ebfa-4a2e-9ba0-bff06738f233
2026-09-22 16:14:10 +02:00

3.5 KiB

Adopting the Emission Cadence contract as a source owner

The Emission Cadence standard is at candidate (document 0.2.0, wire schema version 0.1). Candidate means the contract is specified and machine-checkable. It asserts nothing about adoption: no source has implemented it.

This page is what a source owner needs in order to change that. It asks for a declaration you own, not for a change to your system.

What the contract is for

A source declares how often it emits an event class, so that an observer can tell silence that means nothing happened from silence that means the source stopped. Without a declared cadence, an absent event is unreadable. The standard keeps the two roles apart: the source declares, the observer evaluates, and a declaration is never evidence that the emission actually happens.

What a source owner supplies

  1. A declaration you own, valid against infospace/schemas/emission-cadence.schema.yaml, published in your own repository. It states your source, your stream_id, and one entry per event class you own, in one of the declared forms:
    • expected-rate — at least N events in a window, below which a finding is raised;
    • heartbeat-or-reconciliation — either a heartbeat asserting nothing-to-report, or a comparison between your own counts and the observer's.
  2. The contract digest you implemented against, so the declaration is pinned to a fixed reading of the contract rather than to whatever the canon says later. Current digest: b08b4d95fc4b0bd3 (candidate, document 0.2.0; 972c0b6701d1693f was the 0.1.0 draft text and the same wire schema) (info_tech_canon export-emission-contract <path.tar> regenerates it).
  3. Nothing else. No deployment, no runtime proof, no conformance claim. A declaration is a statement of intended cadence.

Validate before publishing:

info_tech_canon emission-review path/to/your-declaration.yaml

It returns ok plus operational_truth_assessed: false — the command checks your declaration against the schema and deliberately says nothing about whether your system emits at that cadence.

infospace/standards/emission-cadence/examples/qonto-assistant.yaml shows both forms. It is a worked example and explicitly not a source-owned declaration.

What an observer supplies

Separately from any source: a result produced by evaluating a source's declared cadence against events you actually observed, with the comparison references the declaration names. An observer result from the declaring source does not count — the point is that two parties read the same declaration the same way.

What this repository does with it

Validates submitted declarations, records them as adoption evidence, and reports any incompatibility found in the attempt. It does not operate your system, observe your events, or certify your conformance.

Two source-owned declarations from independent owners, plus one third-party observer result, plus a recorded incompatibility or a recorded absence of one, are what the standard's section 10 requires to move from candidate to stable.

What is honestly uncertain

The contract has never been implemented by anyone. The most useful thing a first adopter can report is where it does not fit — a cadence your system has that neither form expresses, a window semantics mismatch, an event class that needs more than one entry. A recorded incompatibility is a successful outcome of this exercise, not a failure of it.