Module batcher 

Types for consolidating, merging, and extracting columnar update collections.

Structs

Chunker

A chunker to transform input data into sorted columns.

ColumnChunker

A chunker that consolidates Column<(D, T, R)> updates into sorted ColumnBody chunks, without round-tripping through columnation.

ColumnMerger

Counterpart to ColInternalMerger (which merges ColumnationStack chunks). Drives the merge batcher with ColumnBody chunks, no columnation detour, by way of the inherent merge_from / extract methods on ColumnBody<(D, T, R)> below.

Traits

MergeChunk

A chunk the merge and extract bodies below write to: the merger’s ColumnBody, and the column pager’s Column until that pager is retired.

Functions

drain_side 🔒

Drain remaining items from one side into result / output.

empty_chunk 🔒

Pop a chunk from stash or allocate a fresh one. Stashed chunks are already cleared via recycle_chunk, so they’re ready for push.

extract

Partition records of source starting at *position into keep (times beyond upper, retained for the next round) and ship (times not beyond upper, sealed into the output batch). Updates frontier with the times of kept records.

gallop 🔒

Advance *lower past every position in [*lower, upper) where cmp returns true.

merge_from

Merge items from sorted inputs into target, advancing positions.

recycle_chunk 🔒

Reset chunk to an empty Typed and push it to stash for reuse.