Feeds, Changes & Deletes
How documents reach the index and how it learns one changed or died: delivery channels, SFTP failure modes, a document identity that survives an update, tombstones, snapshot vs delta feeds, and a restart-safe batch loop.
Last updated
After this section you can
- Choose how each channel reports changes and deletes, and say what its failure looks like
- Give each document an identity that survives an update, so it replaces its chunks and a delete retires them
- Make an ingest loop safe to kill and restart, and prove a change and a delete reached the answers
Keeping the Index True: Feeds, Changes & Deletes
How documents reach the index, and how it learns that one changed or died.
An index stays true only if it learns what changed and what died. Give each document an identity that survives an update (not its hash), replace all of a changed document’s chunks, delete on an explicit record or a verified full snapshot, and make every run safe to restart. This is gate 1: in the index, whole and current.
Content reaches queries only at the switch; a revocation goes live at once (swipe on a phone).