Skip to content

0.1.9

Latest

Choose a tag to compare

@TejasMorbagal TejasMorbagal released this 25 Mar 10:06
· 7 commits to main since this release
  • jupyter_kernel_info is now optional in RecordProperties; workflow configs without a notebook URL no longer require this field.
  • Removed redundant hasattr guard for _last_generator in Publisher.publish().
  • Added automated MkDocs GitHub Pages deployment on release via a dedicated docs.yml workflow.
  • osc_themes values are now automatically lowercased, so 'Land' and 'LAND' are treated the same as 'land', preventing theme validation failures.
  • collection_id is now validated to contain no spaces; a clear error is raised with a hint to use hyphens instead.
  • license_type (dataset) and properties.license (workflow) are now mandatory fields; publishing fails immediately with a descriptive error if either is missing.
  • Variable catalog description now falls back to the title-cased variable ID when neither description nor long_name attrs are present on the zarr variable, preventing null description validation failures.
  • Pull requests opened by deep-code now include a "Generated with deep-code" note in the PR description.
  • stac_catalog_s3_root is now a mandatory field in the dataset config; publishing fails immediately with a descriptive error if it is absent.
  • STAC catalog links in the OSC collection now follow the OSC convention: a via link to the STAC browser URL and a child link to the direct HTTPS catalog URL. The s3:// URL is converted to HTTPS (AWS virtual-hosted style) to satisfy the uri-reference format check in the OSC products schema.
  • Added optional visualisation_link field to the dataset config; when provided, a visualisation link with title "Dataset visualisation" is added to the generated OSC collection.
  • Added optional description field to the dataset config; overrides the description attribute from the Zarr store when set.
  • Added optional osc_project_title field to the dataset config to correctly set the project link title (e.g. "DeepESDL") instead of deriving it from the project ID.
  • Fixed workflow_id not being normalised (slugified) when stored on Publisher, causing spaces in experiment link hrefs and failing uri-reference format validation.
  • Removed redundant via access link from the OSC STAC collection; access is already expressed via typed assets (zarr-data, zarr-consolidated-metadata) on the STAC item.
  • osc_project is now omitted from OscDatasetStacGenerator when not provided, preserving the callee's default instead of passing None.

What's Changed

Full Changelog: v0.1.8...v0.1.9