Skip to contents

Discovers each {prefix}/{version}/catalog.json in the bucket and returns the list release_database.qmd writes as versions.json — one record per release with version, release_date, tables, total_rows, size_mb, plus the archive-policy fields: consolidated (parquet kept indefinitely, from metadata/release_policy.yml) and, when the version's parquet has been removed by scripts/thin_releases.R, retired ({retired_utc, to, reason} read from its retired.json). Both the release notebook and the thinning script call this, so a re-run of the release cannot drop the policy fields.

Usage

build_versions_json(
  bucket,
  prefix = CC_RELEASE_PREFIX,
  consolidated = character(),
  current = NULL
)

Arguments

bucket

GCS bucket

prefix

release prefix (ducklake/releases, or a staging prefix)

consolidated

character vector of consolidated versions

current

optional record for the release being written right now (takes precedence over what the bucket holds for that version)

Value

list of release records, newest first