Skip to content

the_tuva_project v0.17.2

Choose a tag to compare

View all tags
@aneiderhiser aneiderhiser released this 02 Apr 17:07
v0.17.2
cc3b8ce

Highlights

  • Tuva Data Assets on DoltHub: Tuva data assets are now organized as 6 versioned databases on DoltHub: terminology, value sets, reference data, provider data, concept library, and synthetic data. DoltHub is now the release source of truth, while S3, GCS, and Azure Blob remain the distribution layer used by the existing seed-loading macros.
  • Versioned Seed Releases: Package seeds and synthetic integration-test inputs now load from versioned release artifacts that have been copied to S3/Azure/GCS from DoltHub. Releases can be controlled with tuva_seed_version, per-database overrides via tuva_seed_versions, and storage overrides via tuva_seed_buckets.
  • CI Improvements: CI now uses dbt-style command routing instead of alias-style commands. For example, /ci dbt build --full-refresh runs across all 6 warehouses, while /ci snowflake dbt run or /ci redshift dbt run --select core__medical_claim target specific warehouses and selections.
  • Docs in the Tuva Repo: The docs site now lives in the tuva repo under docs, which keeps docs closer to the code and metadata they describe.

Input Layer

  • No net new breaking input-layer contract changes relative to v0.17.1.
  • Input-layer YAML definitions now include clearer mapping_instructions and required_for_data_marts metadata.

Data Mart Updates

  • CMS-HCC Institutional Logic: institutional_status is now derived from long_term_institutional_flag, improving institutional classification in CMS-HCC workflows.

Bug Fixes

  • BigQuery Seed Loading: BigQuery now handles both empty-string nulls and \N null markers in release-backed CSV artifacts.
  • DuckDB Provider Attribution: Fixed a DuckDB planner issue affecting provider attribution workflows.

Miscellaneous

  • integration_tests is Synthetic-Only: integration_tests now runs only against published synthetic datasets, with small as the default size and large also available.
  • Header-Only Synthetic Seed Files: Local CSVs in integration_tests/seeds are now header-only; dbt seed hydrates them from published versioned synthetic artifacts.
  • Seed Publishing Tooling: Added scripts to publish DoltHub-backed seed releases to S3 and mirror them to GCS and Azure.
Assets 2
Loading