- [x] Hashing at multiple points (use HTTP Content-Digest header) - [ ] OI backup - [ ] OI manifest of objects for going offline
Activity
ctengel commentedon Aug 17, 2025
Start with filename based async verify
Long term everything comes with digest and gets verified after write, checksum then also written and then 200 OK
GETs then all include a digest (from digest file) and clients optionally verify
ctengel commentedon Aug 18, 2025
Base "light" metadata
Some readme tips
ctengel commentedon Aug 18, 2025
Careful - not always do we get correct mime from extension
ctengel commentedon Aug 25, 2025
Repr digest
Content Disposition
ctengel commentedon Sep 28, 2025
Hash server logic
On write:
On read:
To fully do this, clients will also require some changes
MIME type, content Disposition etc is a separate issue
OI metadata is sort of also a separate issue
ctengel commentedon Oct 15, 2025
HEAD return repr digest... Get returns both
ctengel commentedon Dec 30, 2025
read/get/head means overriding
send_head()'write/put is minor updates to existing
do_PUTstore checksim on PUT
ctengel commentedon Jan 1, 2026
there is no standard as to which HTTP code to send on digest mismatch, but https://www.ietf.org/archive/id/draft-kleidl-digest-fields-problem-types-01.html#name-mismatching-digest-value seems logical - using 400 instead of 422 or 5xx
ctengel commentedon Jan 1, 2026
better link - https://www.ietf.org/archive/id/draft-ietf-httpapi-digest-fields-problem-types-03.html#name-mismatching-digest-values
ctengel commentedon Jan 2, 2026
https://datatracker.ietf.org/doc/draft-ietf-httpapi-digest-fields-problem-types/03/
ctengel commentedon Jan 4, 2026
allow content-digest header
return hash in ls
Replication hashes and multiple FAPI bugfixes
Script to migrate from old to new checksum style