This release cycle brings an experimental unified schema storage mode to SpiceDB, takes the query planner further toward distributed execution, refreshes the SpiceDB Playground, and updates all of the official client libraries to the latest API.
SpiceDB can now store your whole schema as a single record instead of one record per definition, which makes loading the schema significantly faster.
This is opt-in and experimental. You enable it with the --experimental-schema-mode flag and migrate in stages using rolling restarts. The SpiceDB Operator can step through the stages for you. See the release notes for each stage.
The experimental query planner can now spread work across SpiceDB nodes and cache the results, like the existing engine does. It also simplifies queries before running them.
For a look at what this means for performance, read Query Planner Early Benchmarks. Still experimental and not recommended for production.
Clusters that reference more than one Kubernetes secret no longer run into conflicts, and a crash related to credential configuration is fixed. The operator also supports upgrading to SpiceDB v1.52.0.
The SpiceDB Playground received a UI refresh and a reworked tab system. You can now edit your schema and relationships in the same view, and there are new example schemas to start from. Read the announcement.
The SpiceDB VS Code extension adds YAML validation file support, checks the installed SpiceDB version on launch, and fixes several syntax highlighting issues.