Glossary

Glossary

Canonical data — Research state that must be preserved and is authoritative, such as the main library database and managed research files.

Derived data — Rebuildable data such as thumbnail/search/reference indexes.

Durable operation — A local-first record of user intent stored durably in the browser before the change is reported as saved, then synchronized/reconciled with the server.

Main — The workspace pane/tab that owns the browser URL.

Secondary pane — A visible split-view pane beside/below Main. It has its own TabContext.

Parked tab — An open PRKS tab that is not currently mounted as a visible pane.

TabContext — Per-tab runtime context containing route/page root, async generation, and live resources such as viewers/editors/graphs.

Work — Primary research item: PDF/video/other supported research source plus metadata and relationships.

Work source identity — Aggregate that determines what source a Work actually represents; for video sources it cannot safely be modeled as unrelated independent columns.

Research Note — Work-associated research text that participates in the local-first notes model.

Concept — Structured research idea/category.

Position — Structured claim/position.

Argument / Stance — Structured argumentative relationship with sources/targets.

Read projection / offline cache — Disposable cached representation used to render selected content when the server is unreachable. It is not durable user intent.

Request coordinator — Short-lived online request coordination/cache. It is neither the offline read store nor the durable operation store.

Materialization — Producing/reconciling derived bytes/state from canonical structured data, such as applying the canonical annotation set to managed PDF bytes.

Rollout status — The living document docs/local-first-rollout-status.md that records which local-first families are currently durable/offline-capable.