[filipeforattini]

This PR was opened by the Changesets release GitHub action. When you're ready to do a release, you can merge this and the packages will be published to npm automatically. If you're not ready to do a release yet, that's fine, whenever you add more changesets to main, this PR will be updated.

Releases

@reddb-io/cli@1.25.0

Minor Changes

• #2440 85bd8e8 Thanks @filipeforattini! - Release the KV/SQL performance and durable-write cycle as 1.25: publish
  single-file KV replacements atomically, share durable grouped commits, avoid
  redundant overflow checkpoints and reserve bounded WAL headroom. Preserve both
  durability barriers, fence uncertain publication failures until reopen, and
  restore indexed SQL/KV reads and recovery metadata. Existing storage and wire
  formats remain unchanged. Full snapshots still cause pauses; measured results
  and upgrade boundaries are documented in docs/release/v1.25.0.md.

Patch Changes

• #2431 51d4a6d Thanks @filipeforattini! - Bound materialization for simple table LIMIT/OFFSET windows and avoid reserving
  memory for skipped rows or arbitrarily large requested limits.

• #2439 6eb66dc Thanks @filipeforattini! - Enforce the configured read-result byte ceiling with bounded JSON serialization,
  add materialized-read concurrency admission, and apply cooperative query deadlines
  to existing read work checkpoints. Output streams charge cumulative payload bytes.
  Mutation acknowledgements remain outside read-result limits; operator/cgroup and
  materialization limits remain necessary for memory and non-cooperative work.

• #2428 bd94c3d Thanks @filipeforattini! - Use segment key indexes for KV point reads instead of materializing entire collections, while retaining MVCC version selection and recovery behavior.

• #2432 a3af4d6 Thanks @filipeforattini! - Extract KV read values directly from indexed candidates instead of cloning whole rows, preserving snapshot visibility and mutation behavior.

• #2433 dd8fcc1 Thanks @filipeforattini! - Retain active statement snapshot horizons through query completion so concurrent
  VACUUM cannot reclaim versions still needed by ordinary or nested reads.

• #2441 9ea24de Thanks @filipeforattini! - Update rustls-webpki to 0.103.16 and webpki-roots to 1.0.10 across the workspace,
  Python driver and Node addon lockfiles. Reject distrusted emSign intermediate
  keys and remove the affected bundled root, addressing RUSTSEC-2026-0382 and
  RUSTSEC-2026-0381. Connections using affected certificate chains may now fail
  verification; replace the chain rather than disabling verification.

• #2429 2169185 Thanks @filipeforattini! - Preserve row versions required by snapshot transactions that excluded an older
  writer. VACUUM now retains that snapshot's horizon after the writer commits and
  releases it when the reader commits or rolls back.

@reddb-io/client-bun@1.25.0

@reddb-io/sdk@1.25.0

@reddb-io/client@1.25.0

@reddb-io/mcp@1.25.0

@reddb-io/internal-asset-fetcher@1.25.0

@reddb-io/internal-bin-resolver@1.25.0