Files
reuse-surface/history/2026-07-07-inter-hub-retirement.md
tegwick 9b02e54bb6
Some checks failed
ci / validate-registry (push) Has been cancelled
Document inter-hub retirement and maturity reassessment
inter-hub is retired (Haskell/IHP build-chain problems), superseded by
core-hub (Python/FastAPI/Postgres). Its capability entry was updated
directly in ~/inter-hub (commit cffdf9e): D3/A1/C1/R0 -> D5/A1/C3/R0,
status: deprecated, relations.related_to -> capability.infotech.core-hub.

core-hub itself is not yet added to reuse-surface's roster/federation
sources -- it postdates the WP-0017 coverage sweep; noted as a follow-up.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
2026-07-07 01:45:00 +02:00

3.5 KiB
Raw Blame History

inter-hub retirement and maturity reassessment

Follow-up to the REUSE-WP-0017 coverage sweep, requested by Bernd after reviewing history/2026-07-06-t04-review-summary.md and the READMEs written for inter-hub/open-reuse/vantage-point.

Decision

inter-hub (the Haskell/IHP Interaction Hub Framework implementation) is retired due to persistent build-chain problems. The same framework ambition is being rebuilt on a more convenient technology stack as core-hub (Python/FastAPI/Postgres, contract-first). This is an owner decision (Bernd), not a reuse-surface judgment call.

core-hub already exists as a repo (~/core-hub, first commit 2026-06-27) with its own self-authored capability entry (capability.infotech.core-hub, D3/A1/C1/R1) — it was created after the last workstation roster snapshot (2026-06-16) and REUSE-WP-0017's coverage sweep, so it isn't yet part of reuse-surface's federation sources or the 61-repo roster. core-hub/SCOPE.md explicitly frames itself as the "Gen 3" successor to inter-hub ("Gen 2") and state-hub ("Gen 1"), and already scopes in Inter-Hub /api/v2 compatibility and a data-migration plan.

Maturity reassessment for capability.infotech.interaction-hub-framework

The original REUSE-WP-0017-T04 first-pass entry (2026-07-06) rated this D3/A1/C1/R0 with availability confidence explicitly flagged low ("not independently verified in this sweep"). A deeper look at SCOPE.md's own per-phase log, prompted by writing inter-hub's real README, showed Phases 09 complete: real Web/Controller/ modules (governance, requirements, hub capability manifests, federated policy overlays, deployment records, agent registrations, webhooks), a Test/ suite, and a versioned external API (/api/v2) with OpenAPI, OAuth client credentials, generated SDKs, and rate limiting — a substantial working reference implementation, not a paper spec.

New vector: D5 / A1 / C3 / R0, status: deprecated.

Axis Change Why
Discovery D3 → D5 Undersold on first pass; Phases 0-9 are genuinely documented and delivered
Availability A1 → A1 (held) Deliberately not raised — the Haskell/IHP build chain is the specific reason for retirement. Feature completeness and build/deploy reliability are different axes; this is exactly the case where they diverge
Completeness C1 → C3 Nine implementation phases delivered against the original spec's traceability chain
Reliability R0 → R0 (held) The retirement itself is the reliability signal — a feature-complete implementation that could not be kept dependably buildable. Recorded as an incident, not glossed over

relations.related_to now points at capability.infotech.core-hub. consumer_guidance reframed: recommended only for historical reference/prior-art study; explicitly not recommended for new integration work.

Committed and pushed directly in ~/inter-hub (single-repo change, not a mass push): commit cffdf9e.

Not done (out of scope for this follow-up)

  • core-hub is not yet added to reuse-surface's local-repo-roster.yaml or registry/federation/sources.yaml — it wasn't part of the original 61-repo coverage sweep and adding it means a fresh registration pass (publish-check, hub registration) of its own. Worth doing as a small follow-up so the related_to relation resolves inside reuse-surface's own federated index, not just as a prose pointer.
  • core-hub's own capability entry (D3/A1/C1/R1) was not touched — it's self-authored and self-governed by that repo, not part of this reassessment.