API v1 Requirements about spicedb HOT 2 CLOSED

Some other potential improvements:

• delete API which takes a list of tupleset filters as a mirror to the Read API
• definitely remove the user oneof where we used to allow integer user IDs

from spicedb.

Support for schema language instead of namespace configs

Done in v1alpha1 and supported further in v1, where schema reads now return the full set of schema and schema writes will now remove any unreferenced types (i.e. it is a "full" write)

All API calls are consistent by default (not done)
Zookies optionally can be provided to safely allow flexibility in consistency

The decision was made to keep existing "good enough" behavior to keep the performance benefits, but to also allow explicit configuration of the consistency level required on a request via the https://github.com/authzed/api/blob/main/authzed/api/v1/permission_service.proto#L50 message

Subject References should treat a lack of relation as "ellipsis"

Explicitly moved to an optional relation on subjects which, when not specified, indicates the relationship is to the subject as a whole (equivalent to ... in V0).

delete API which takes a list of tupleset filters as a mirror to the Read API

Done, and also applied to the Write API for preconditions

definitely remove the user oneof where we used to allow integer user IDs

Done, as integer IDs were already disallowed in V0

from spicedb.