# Changelog
All notable changes to this project will be documented in this file.
The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/),
and this project adheres to the [Haskell Package Versioning Policy](https://pvp.haskell.org/).
## 0.2.0.0 — 2026-05-07
### Changed (BREAKING)
- Renamed modules and functions to use streamly's vocabulary instead of
conduit's. The functions return Streamly `Stream`s and `Fold`s, so the
modules are now named `Kafka.Streamly.Stream` and `Kafka.Streamly.Fold`,
and the function families are `kafkaStream*` and `kafkaFold*`.
| Old (0.1.0.0) | New (0.2.0.0) |
|--------------------------------|------------------------------|
| `Kafka.Streamly.Source` | `Kafka.Streamly.Stream` |
| `Kafka.Streamly.Sink` | `Kafka.Streamly.Fold` |
| `kafkaSource` | `kafkaStream` |
| `kafkaSourceAutoClose` | `kafkaStreamAutoClose` |
| `kafkaSourceNoClose` | `kafkaStreamNoClose` |
| `kafkaSink` | `kafkaFold` |
| `kafkaBatchSink` | `kafkaBatchFold` |
No backward-compatible re-exports are provided. Callers porting from
0.1.0.0 should mechanically substitute identifiers per the table.
## 0.1.0.0 — 2026-04-17
Initial release.
### Added
- `Kafka.Streamly.Source` — three consumer stream variants (`kafkaSource`,
`kafkaSourceAutoClose`, `kafkaSourceNoClose`) for different resource-ownership
models; error predicates (`isFatal`, `isPollTimeout`, `isPartitionEOF`);
error filters (`skipNonFatal`, `skipNonFatalExcept`); and value-mapping
helpers built on `Bifunctor` (`mapFirst`, `mapValue`, `bimapValue`).
- `Kafka.Streamly.Sink` — two producer fold variants (`kafkaSink` for
per-record sends, `kafkaBatchSink` for `[ProducerRecord]` batches) and a
`withKafkaProducer` bracket helper that flushes and closes on exit.
- `Kafka.Streamly.Combinators` — `batchByOrFlush` and `batchByOrFlushEither`
for size-bounded batching with explicit flush, and `throwLeft` /
`throwLeftSatisfy` for raising error values as exceptions.
- Module-level Haddock with worked examples and `@since 0.1.0.0` tags on
every exported identifier.
### Fixed
- `withKafkaProducer` no longer flushes twice on teardown
(`Kafka.Producer.closeProducer` already flushes internally).
### Changed
- `batchByOrFlush` and `batchByOrFlushEither` now reject non-positive
`BatchSize` with an explicit error rather than silently emitting
singleton batches.
- Value-mapping helpers pruned: the nine `sequenceValue*`, `traverseValue*`,
and `bitraverseValue*`/`bisequenceValue` variants have been removed.
Callers who relied on them can use `fmap`/`bimap`/`traverse`/`bitraverse`
inline. `mapFirst`, `mapValue`, and `bimapValue` remain, re-documented as
`Bifunctor`/`Functor` lifts.
### Tests
- Initial pure test suite covering error predicates (`isFatal`,
`isPollTimeout`, `isPartitionEOF`), error filters (`skipNonFatal`,
`skipNonFatalExcept`), batching combinators (`batchByOrFlush`,
`batchByOrFlushEither`), and exception combinators (`throwLeft`,
`throwLeftSatisfy`). Uses `tasty` + `tasty-hunit` + `tasty-quickcheck`.
Run with `cabal test`.