Major Description
Currently backups read and write files (either locally, or across a network), but nothing is done to make sure that a backup isn't backing up index files that are already corrupted. Or to make sure that corruption doesn't occur in transit. Backup/restore should ahve some protections to ensure that files that are already corrupted aren't persisted. We can do this by computing a checksum of the file as it's being read/written and comparing that to values stored in the index files already.
See SIP-12 and the corresponding discussion thread for more details.