Three shapes of data

Super Artifacts holds data in three shapes. They differ in who writes the rules, and that decides everything else about them.

Responses an artifact collects

When an artifact takes in what visitors type, its agent declares a policy at publish: the fields, who may read each answer back, how long to keep them, which ones are about a person. The artifact only declares. The platform enforces, because code inside an artifact is the one thing that cannot be trusted to enforce its own rules.

A policy is additive. A later version may add fields, but a field name keeps its type for good, because the answers already collected would no longer fit; to change one, the agent adds a new field that replaces it. An artifact with no policy collects nothing.

Collections you edit

A collection is a table a person authors in the app: fields, entries, links between collections, and some fields kept encrypted. You edit the shape freely, because you know what it was meant to hold. A collection can have more kinds of field than a policy can, since nobody's prompt gets longer when one is added.

Datasets across artifacts

A dataset is the name an artifact's responses are filed under. Datasets that share a series can be queried as one, so a question can span a survey and the one that replaced it. A series says what may be read together; it says nothing about who may read it.

Where each reads back

All three appear on the Data page. An agent reads responses with super_data and super_data_query, and collection entries with super_entries. Sizes and counts are in Limits.

Next: Limits