Changelog

All notable changes to this project will be documented in this file.

This client follows Semantic Versioning and targets the stable chunkdb 1.x protocol; see the engine's compatibility policy.

1.1.0 - 2026-07-18

chunkScan(limit, cursor?), chunkRange(cx0, cy0, cx1, cy1), and chunkRadius(cx, cy, radiusChunks) for world streaming (paginated populated-chunk enumeration plus bounded rectangular and radius reads)
chunkVersion(cx, cy), chunkCompareAndSet(...), and chunkBatch(...) for optimistic concurrency on a single chunk
walFlush() explicit durability barrier
metrics() Prometheus text-format runtime metrics
chunkbinCompressed(cx, cy) / chunkbinStateCompressed(cx, cy) using the server's zrle codec, plus exported zrleCompress/zrleDecompress
ChunkPool now mirrors every new ChunkClient operation (world reads, concurrency primitives, compressed reads, walFlush, metrics)

Requires a chunkdb server that implements these additive commands; all existing methods keep working against older 1.x servers.

First stable release of @chunkdb/client, aligned with stable chunkdb 1.0.0.

mset(blocks) / mget(blocks) — batch multi-block write/read in a single round-trip (protocol MSET/MGET with *N array reply)
request pipelining: pipelineDepth client option allows multiple in-flight requests per connection (default 1); mset/mget also exposed on the pool
ArrayFrame protocol frame type and *N array parsing

internal pending-request model reworked from a single in-flight slot to a FIFO queue to support pipelining

reject command parts containing CR/LF before serialization (request-injection guard)