Data Feeds is in early access. Contact us to set up
a subscription; the API and sync client described here open to early access
customers first.
- a snapshot: the whole dataset as of that version
- a changes file: the records added, changed and removed since the version before it
- its schema
Listing versions
Getting a version’s files
GET /v1/datasets/{id}/versions/{version} returns the version with signed links to its files. Links expire after files.expiresAt; ask again for fresh ones. Pass ?format=csv for CSV; GeoParquet is the default.
sha256 after download.
Applying changes
A changes file holds one row per record that changed, with the envelope columns and the record’s columns as of this version:
Apply versions in order. If you have version 40 and the current version is 42, apply 41’s changes and then 42’s.
For small versions,
GET /v1/datasets/{id}/versions/{version}/changes returns the same changes as JSON pages, without downloading a file.
When to reload instead
Reload a dataset from a version’s snapshot when:- the version has
resyncRequired: true(its key columns changed, so record ids did too) - you receive
dataset.resync_required - the version you need is past retention (
410)
Retention
A subscriber who stays within 90 days can always catch up by changes alone.