Struct CompactionExecutor 

pub struct CompactionExecutor { /* private fields */ }

Executes compaction plans: reads N small files, merges them into a single AI-Lake file with a rebuilt index, and commits to the catalog.

The index algorithm is chosen via CompactionIndexStrategy (default: Auto, which detects GPU / CPU cores at compaction time — the same heuristic used by write_batch_auto).

Implementations

impl CompactionExecutor

pub fn new(store: Arc<dyn Store>, policy: VectorStoragePolicy) -> Self

pub fn with_index_strategy(self, strategy: CompactionIndexStrategy) -> Self

Override the default (Auto) index strategy for this executor.

pub async fn compact( &self, files: &[DataFileEntry], output_path: &str, ) -> AilakeResult<DataFileEntry>

Merge files into a single new file at output_path. Returns the DataFileEntry for the merged file.

pub async fn run( &self, planner: &CompactionPlanner, table: &TableIdent, catalog: Arc<dyn CatalogProvider>, output_prefix: &str, ) -> AilakeResult<Option<DataFileEntry>>

Full compaction workflow: plan, compact, drop old files from catalog, commit.