[Epic] Deliver the governed datasource catalogue and lifecycle #1

Closed
opened 2026-07-28 12:48:18 +02:00 by zemion · 3 comments
Owner

Outcome

Provide one provider-neutral, tenant-governed data layer that Dataflow,
Workflow, Reporting, Risk Compliance, and domain modules can consume without
importing connector implementations or retaining credentials.

Boundary

  • Datasources owns source identity, visibility, live/cached/static mode,
    staging, immutable materializations, frozen states, schema/fingerprints,
    provenance, and bounded reads.
  • Connectors owns external protocols, endpoints, credentials, discovery,
    source-side filtering/pushdown, and provider health.
  • Dataflow owns transformations and lineage.
  • Workflow owns orchestration, approvals, retries, and module handoffs.

First Slice

  • Package, module manifest, permissions, migration, uninstall guard, and WebUI route.
  • Tenant-isolated live, cached, and static datasource catalogue.
  • Bounded JSON/CSV staging and promotion.
  • Append-only materializations, current state, and explicit frozen states.
  • Schema versions, content fingerprints, provenance, and bounded preview.
  • Connector origin adapter and Dataflow capability consumer.
  • Focused composition test from Connector through Datasources to Dataflow.
  • Production storage and concurrency hardening.
  • Quality, schema compatibility, approval, retention, and policy gates.
  • Module-output publication and Workflow lifecycle integration.

Architecture: https://git.add-ideas.de/GovOPlaN/govoplan/src/branch/main/docs/DATASOURCE_AND_DEFINITION_GRAPH_ARCHITECTURE.md

Architecture reconciliation (2026-07-31)

The implemented lifecycle remains the foundation. Typed institutional catalogue metadata and dependency impact are the child slice GovOPlaN/govoplan-datasources#6. This does not move connector profiles/credentials, transformations, or report execution into Datasources.

## Outcome Provide one provider-neutral, tenant-governed data layer that Dataflow, Workflow, Reporting, Risk Compliance, and domain modules can consume without importing connector implementations or retaining credentials. ## Boundary - Datasources owns source identity, visibility, live/cached/static mode, staging, immutable materializations, frozen states, schema/fingerprints, provenance, and bounded reads. - Connectors owns external protocols, endpoints, credentials, discovery, source-side filtering/pushdown, and provider health. - Dataflow owns transformations and lineage. - Workflow owns orchestration, approvals, retries, and module handoffs. ## First Slice - [x] Package, module manifest, permissions, migration, uninstall guard, and WebUI route. - [x] Tenant-isolated live, cached, and static datasource catalogue. - [x] Bounded JSON/CSV staging and promotion. - [x] Append-only materializations, current state, and explicit frozen states. - [x] Schema versions, content fingerprints, provenance, and bounded preview. - [x] Connector origin adapter and Dataflow capability consumer. - [x] Focused composition test from Connector through Datasources to Dataflow. - [x] Production storage and concurrency hardening. - [x] Quality, schema compatibility, approval, retention, and policy gates. - [x] Module-output publication and Workflow lifecycle integration. Architecture: https://git.add-ideas.de/GovOPlaN/govoplan/src/branch/main/docs/DATASOURCE_AND_DEFINITION_GRAPH_ARCHITECTURE.md ## Architecture reconciliation (2026-07-31) The implemented lifecycle remains the foundation. Typed institutional catalogue metadata and dependency impact are the child slice [GovOPlaN/govoplan-datasources#6](https://git.add-ideas.de/GovOPlaN/govoplan-datasources/issues/6). This does not move connector profiles/credentials, transformations, or report execution into Datasources. <!-- codex-institutional-governance-datasources-2026-07-31 -->
Author
Owner

Codex State: progress

Pushed 492306f: Datasources now contributes a permission-aware native PostgreSQL Search source. Backfill is tenant-bounded and current/non-deleted, emits only safe catalogue metadata, excludes rows/schema/provider references/credentials/provenance, rechecks current catalogue authorization, and deep-links into the selected datasource.

All 31 Datasources tests pass (two optional PostgreSQL integration cases skipped without the test URL), with Ruff and WebUI typechecking green. The epic remains open for its production hardening, quality/policy gates, and module-output/Workflow lifecycle criteria.

## Codex State: progress Pushed `492306f`: Datasources now contributes a permission-aware native PostgreSQL Search source. Backfill is tenant-bounded and current/non-deleted, emits only safe catalogue metadata, excludes rows/schema/provider references/credentials/provenance, rechecks current catalogue authorization, and deep-links into the selected datasource. All 31 Datasources tests pass (two optional PostgreSQL integration cases skipped without the test URL), with Ruff and WebUI typechecking green. The epic remains open for its production hardening, quality/policy gates, and module-output/Workflow lifecycle criteria.
Author
Owner

Progress: child #2 is complete and released in v0.1.21. Datasources now covers quality/schema gates, optional attributable approval for stages and cached refreshes, explicit preview/apply retention for stages/materializations/frozen evidence, and hash-chained lifecycle evidence. Full focused workspace verification passed. Parent remains open for its remaining child slices.

Progress: child #2 is complete and released in v0.1.21. Datasources now covers quality/schema gates, optional attributable approval for stages and cached refreshes, explicit preview/apply retention for stages/materializations/frozen evidence, and hash-chained lifecycle evidence. Full focused workspace verification passed. Parent remains open for its remaining child slices.
Author
Owner

Epic completion verified after #2 closed. All child slices #2-#8 are complete: lifecycle quality/approval/retention, immutable output publication and terminal handoffs, concurrency/storage hardening, policy-aware visibility, institutional catalogue semantics, interface-pattern migration, and DSAR coverage. The Connector -> Datasources -> pinned Dataflow composition and full focused workspace gate pass on released Datasources v0.1.21.

Epic completion verified after #2 closed. All child slices #2-#8 are complete: lifecycle quality/approval/retention, immutable output publication and terminal handoffs, concurrency/storage hardening, policy-aware visibility, institutional catalogue semantics, interface-pattern migration, and DSAR coverage. The Connector -> Datasources -> pinned Dataflow composition and full focused workspace gate pass on released Datasources v0.1.21.
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: GovOPlaN/govoplan-datasources#1