Ratify an ontology-first schema-change and migration process
Context and Problem Statement
The 5 October call distinguishes the existing schema-first GitHub pull-request process from the intended future flow: ontology proposal, human review and approval, then generation and release of a schema. Paul asked Henrik to use the outstanding PRs as a worked example and training package. Treating their content as hypothetical approved input for an exercise was not permission to approve or merge them.
Paul said other knowledgeable technical subgroup members should approve the current PRs. Henrik should not click approve without the necessary context. The new process needs subgroup ownership and sign-off so it does not depend on Henrik continuing as modeller. Compatibility and cutover must also be agreed by the wider technology team; the call did not settle a migration design or transition duration.
Rick Saville’s 8 October email proposes comparing LMS’s modular PDTF schemas against ontology outputs. This is concrete evidence for the transition exercise, not proof that either the modular schema or a generated replacement has been adopted.
Considered Options
- Continue schema-first changes indefinitely.
- Switch immediately to generated schemas on the modeller’s authority.
- Demonstrate and ratify ontology-first changes, with explicit transition controls.
Decision Outcome
Propose the third option. Henrik prepares the workflow and a worked example using identified PRs: source request, semantic impact, proposed ontology change, review, draft schema generation, validation, independent approval and versioned release. Show the exact artefacts and approval boundaries; draft generation is not release.
The internal technical subgroup should own the process and record its approval before cutover. Paul convenes the review and obtains current PR reviews from the appropriate members. Existing PR handling remains under the existing process until an authorised transition decision says otherwise.
Henrik and Rick should compare modular schema outputs and ontology-derived outputs, recording structural differences, semantic coverage, constraints and mappings. The technology team then determines compatibility obligations, any adapters, coexistence period, deprecation criteria and release responsibility. Do not promise byte-identical schemas, automatic equivalence or a painless migration without evidence. Preserve ADR-0077’s distinction between the third-party PDTF input, its technical derivation and the collaboratively authored SPDTF candidate.
Consequences
- Good, because the proposed process is understandable and owned by its human reviewers.
- Good, because real PRs and LMS material make compatibility differences testable.
- Bad, because generation and migration need evidence beyond a convincing demonstration.
- Neutral, because this proposal does not approve current PRs or retire existing schemas.
Confirmation
No generation implementation, PR approval, merge, migration or process ratification is claimed here. Rick offered Tuesday or Friday the following week, 09:00–11:00; Henrik replied that Tuesday works. The read thread does not confirm a time.
| Owner | Outstanding action | Completion evidence |
|---|---|---|
| Henrik | Prepare process documentation and a PR-based demonstration/training package. | Versioned example reviewed by the technical subgroup. |
| Henrik + Rick | Confirm a meeting time and compare the modular PDTF with ontology outputs. | Agreed time, identified input versions and recorded differences. |
| Paul + appropriate reviewers | Review existing PRs and convene process sign-off. | PR reviews and a recorded process decision. |
| Technology team, with Henrik supplying analysis | Choose and approve the compatibility/cutover approach. | Tested migration proposal and explicit approval. |
Comments
Loading comments…
Sign in to post a comment