Each file-level object should have a unique identifier
Closed this issue · 2 comments
In the new_manifest_proposal.json file, it would be useful if each file had assigned a unique identifier, or if a specified formula was given to calculate one (such as the MD5 of the collection, path, and filename).
This identifier would be useful for indexing.
I don't think it would be sensible to create and maintain an id for the purpose of indexing since one can be generated in other ways. (trivially: depositor/collection/path/filename as you point out)
The current proposal emphasizes packages and unique package identifiers as the key to connecting data (packages) in archival storage and metadata (about packages and their contents) in ArchivesSpace. The identifier for a file-level object will be a combination of the package_id and path/filename for the file.