keksmd opened a new issue, #978: URL: https://github.com/apache/incubator-graphar/issues/978
### Describe the enhancement requested Add the GraphAr read facade: the first vertical of #947, which turns a GraphAr dataset into vertex, edge, and neighbor streams without JNI. Component(s): Java ## Scope - Add `graphar-reader` with a `GraphReader` facade over `graphar-info`, `graphar-core`, and the `PhysicalReader` contract from `graphar-io-api`. - Vertex property scans by chunk, with projection. - `ordered_by_source` adjacency: resolve `vertexId -> offset chunk -> minimal row range -> destinations` (`neighbors`), plus streaming `scanEdges` over the whole edge type. - Edge property cursors aligned with the topology rows. - Tests against the canonical `testing/ldbc_sample` Parquet fixtures, compared with the C++ reader's results. ## Non-goals No graph algorithms, CSR materialization, traversal, snapshots, other adjacency layouts, filters beyond what `ReadRequest` already carries, or concrete format types in the reader API. The reader talks to `graphar-io-api` only; `graphar-io-parquet` is a test dependency. ## Compatibility and acceptance Reads datasets produced by the C++ and Spark implementations unchanged. A `neighbors` call on one vertex reads only the offset entry and the destination row range it needs, verified by counting storage reads. Java 11 compatible. ## Dependency order Depends on the Parquet backend (#964, PR #975). The PR will be opened after #975 lands; the writer (#973) and validator (#974) follow it. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
