We intentionally separate oid types from hash types; a hash is a generic hunk of bytes, an object id has meaning and backs an object on disk. As a result of this separation, we need a 1:1 mapping.
