Core concepts

summa-core owns storage, indexing, scoring and immutable segment publication. The server, broker, clients and WASM bindings expose that shared engine. Summa 2 uses its own Rust search engine; the original Tantivy-based API does not describe its schema or index format.

An index has a schema, a set of immutable segments and a writer. Indexing stages documents. A commit publishes a new searchable generation; merges combine compatible segments while preserving their stored semantics. Document addresses contain a segment ID and a segment-local document ID.

Use the server for persistent indexes accessed over gRPC, or the browser API for writable local indexes and read-only HTTP/IPFS readers. Storage capabilities depend on the entry point; an old IndexEngine configuration cannot be passed to the new server.


Table of contents


This site uses Just the Docs, a documentation theme for Jekyll.