Changelog

Release log

All versions and changes, newest first.

v1.0

General availability

  • feat Stable API contract: v0.x breaking changes addressed; feldera-client SDK 1.0 released on PyPI and npm.
  • feat Cloud web UI GA: edit pipelines, configure connectors, inspect live output, and manage team access.
  • feat Point-in-time replay: restore state from any logged checkpoint within the retention window.
  • feat Team permissions: owner, editor, read-only roles with per-pipeline access grants.
  • perf Engine operator startup is 60% faster for pipelines with more than 8 join operators.
  • fix Rolling window bounds now correctly handle DST transitions across timezone-aware timestamp columns.
v0.8

Snapshot isolation, Redpanda connector

  • feat Snapshot isolation: checkpoint pipeline state on a configurable schedule and replay from any snapshot.
  • feat Redpanda connector: native API, no Kafka emulation layer.
  • feat Schema registry support for Confluent-compatible Avro and Protobuf input streams.
  • perf JOIN operator memory use cut 35% for high-cardinality key joins.
  • fix Corrected incremental aggregation recomputation for out-of-order late events in TUMBLE windows.
v0.5

S3 connector, window updates

  • feat Amazon S3 source connector: ingest partitioned S3 prefixes incrementally, supporting Parquet and JSON.
  • feat Added HOP (sliding) windows alongside existing TUMBLE windows.
  • feat REST API output connector with configurable key field and optional auth token passthrough.
  • perf Incremental GROUP BY on string keys is 2.4x faster after a hash function update.
  • fix Pipeline restart now retains the first event batch after bootstrapping.
v0.4

MySQL binlog, multi-source joins

  • feat MySQL 8.x binlog CDC connector with row-based replication and GTID offset tracking.
  • feat Multi-source incremental JOIN: join four sources in one query, each using a different connector type.
  • feat Feature API: fetch an individual row by key without reading the full output stream.
  • perf Postgres CDC throughput rose 3x for large-schema tables with many nullable columns.
  • fix CASE WHEN inside GROUP BY aggregations now emits correct deltas on retraction.
v0.3

First public release

  • feat Incremental SQL engine: standard SQL SELECT/GROUP BY/JOIN maintained against streaming input.
  • feat Kafka and Redpanda source connector with JSON schema inference.
  • feat Postgres CDC via logical replication.
  • feat CLI tooling: feldera pipeline deploy, feldera pipeline status, feldera pipeline pause/resume.
  • feat Apache 2.0 release: self-hosted Docker Compose and Kubernetes Helm chart.

Stay updated

Track development

Releases arrive every 6-8 weeks. See the engineering blog for context on major updates.

Engineering blog