There’s a problem I’ve been stewing over.
What’s the best way to organize collections of metadata into repos?
Solution 1: keep the metadata file in the repo with the book. People who know the book are best positioned to accept pull requests.
Solution 2: keep the metadata files in a separate repo just for metadata. Metadata specialists are better positioned to make sure the metadata is clean and uniform.
New solution: Dual metadata. keep the metadata files in both book and metadata repos. Have a bot that syncs them. Have a flexible mechanism to let maintainers have sync commits go directly to master or to require approval.
Byproduct: Use same sync mechanism to allow delegation/syncing of entity metadata, for example metadata about authors of a book. (like foreign keys in a relational db)