v0.5.0: matching flat schemas for nmdc-schema 11.23.0 and 11.24.0

This release ships two flat schemas, each generated from an exact nmdc-schema release, so a consumer can export with the one that matches the source data it reads. Both use projection 1.3.0.

• schema/nmdc_schema_flattened.yaml: nmdc-schema 11.24.0, version 11.24.0+flat.1.3.0, 61 tables. This is the default.
• schema/compat/11.23.0/nmdc_schema_flattened.yaml: nmdc-schema 11.23.0, version 11.23.0+flat.1.3.0, 60 tables. This matches production, which still runs 11.23.0.

Consumers pick a schema by their installed nmdc-schema version with nmdc_lakehouse_schema.artifacts.flat_schema_resource(version("nmdc-schema")). Any other version raises an error; there is no nearest-version fallback. See https://microbiomedata.github.io/nmdc-lakehouse-schema/source-versions/.

Also in this release:

• Substances nested in mobile-phase segments are preserved in their own helper tables (#22, fixing #21).
• CI regenerates both schemas from their pinned source releases and fails if either checked-in file differs.
• The PyPI page links to the docs, repository, issues and releases (#25).

This release does not migrate MongoDB or produce a production export. The lakehouse adopts it in microbiomedata/nmdc-lakehouse#348.

v0.4.0: NMDC 11.24.0 with TextValue parent columns

This release adopts the latest tagged source schema, nmdc-schema 11.24.0, and ships target schema 11.24.0+flat.1.2.0 with 59 tables (19 primary and 40 helpers).

• TextValue slots become strings or string arrays on their containing records; the 41 former TextValue-only helper tables are removed.
• Person and Organization credit fields are projected through applies_to_agent; data generation gains a credit-association helper table.
• Retired principal-investigator columns and collection_date_inc are removed to match the source release.
• Collection and non-TextValue child record types remain available.
• Wheel and source-archive contents are verified against the reproducible, digest-checked canonical schema before publication.

Validation on merged PR #20: 157 tests, distribution verification, documentation generation/build, and CI on Python 3.10–3.14 passed. Copilot reported no findings. See PRs #15, #18, #19, and #20 for implementation, corrections, and documentation.

Production rollout remains gated. The read-only 2026-09-22 preflight found populated nested substances that projection 1.2.0 does not preserve (#21), and production records still using old investigator/person fields (microbiomedata/nmdc-lakehouse#347). This release does not migrate MongoDB or certify a complete production export. Consumers must use the matching source/schema-package pair and resolve those blockers before rollout.

The lakehouse adoption is microbiomedata/nmdc-lakehouse#340. Existing snapshots retain their original schema identities; no data is relabeled or rewritten by installing this package.

v0.3.0

What's Changed

• fix(docs): deploy the flattened schema, not the copier boilerplate by @sierra-moxon in #8
• Ship the flattened-schema artifact as the repo's product (nmdc-schema 11.23.0) by @sierra-moxon in #10

Full Changelog: v0.2.0...v0.3.0

v0.2.0

What's Changed

• Make the flattened schema purely structural (drop producer identity) by @sierra-moxon in #6

Full Changelog: v0.1.0...v0.2.0

v0.1.0

What's Changed

• Adopt flatten/generate/diff engine as the schema product; invert ETL dependency by @sierra-moxon in #5
• Bump the github-actions group with 7 updates by @dependabot[bot] in #1

New Contributors

• @sierra-moxon made their first contribution in #5
• @dependabot[bot] made their first contribution in #1

Full Changelog: https://github.com/microbiomedata/nmdc-lakehouse-schema/commits/v0.1.0