Skip to content

Remote datasets

GeoCase keeps its bundled core intentionally small. Larger or richer cases can be stored remotely and fetched on demand.

Status: manifests in v1.0, transport in v1.1

v1.0 ships the metadata half of this design. Remote cases are described by manifests and are discoverable by id, but GeoCase does not yet download, cache, or unpack them — asking for a remote case raises a clear error telling you where the artifact lives. The download and cache layer described below is planned for v1.1. See the roadmap for the current scope.

Why remote storage exists

Some cases are useful but not suitable for packaging:

  • larger rasters
  • multi-file archives
  • realistic scene bundles
  • optional public downloads
  • organization-specific private cases

Storage classes

GeoCase recognizes three storage classes:

  • bundled
  • remote
  • private

Remote manifests

Remote datasets are described by manifest files. A manifest maps:

  • case id
  • version
  • relative path
  • checksum
  • size
  • storage backend

GeoCase can use the manifest to:

  • locate a remote case
  • download it into cache
  • verify checksum
  • expose the case through the same API as bundled cases
  1. define the case metadata
  2. publish the archive to object storage or static hosting
  3. add the case to a manifest
  4. fetch on demand during testing

Private cases

Organizations can later maintain their own private manifests and point GeoCase to them without publishing the underlying datasets.