|
1 | 1 | # Version changelog |
2 | 2 |
|
| 3 | +## Release v0.2.0 |
| 4 | + |
| 5 | +#### Native Rust Backend (JNI Migration) |
| 6 | +- The SDK now uses JNI (Java Native Interface) to call the Zerobus Rust SDK instead of pure Java gRPC calls |
| 7 | +- Native library is automatically loaded from the classpath or system library path |
| 8 | + |
| 9 | +#### New APIs |
| 10 | + |
| 11 | +**Offset-Based Ingestion API** - Preferred alternative to CompletableFuture-based API: |
| 12 | +- `ZerobusStream.ingestRecordOffset(IngestableRecord)` - Returns offset immediately without future allocation |
| 13 | +- `ZerobusStream.ingestRecordsOffset(Iterable)` - Batch ingestion returning `Optional<Long>` (empty for empty batch) |
| 14 | +- `ZerobusStream.waitForOffset(long)` - Block until specific offset is acknowledged |
| 15 | + |
| 16 | +**JSON Record Support**: |
| 17 | +- `IngestableRecord` interface - Unified interface for all record types |
| 18 | +- `JsonRecord` class - JSON string wrapper implementing IngestableRecord |
| 19 | +- `ProtoRecord<T>` class - Protocol Buffer wrapper implementing IngestableRecord |
| 20 | +- `RecordType` enum - Specifies stream serialization format (`PROTO` or `JSON`) |
| 21 | +- `StreamConfigurationOptions.setRecordType(RecordType)` - Configure stream for JSON or Proto records |
| 22 | +- Both record types work with `ingestRecord()` and `ingestRecordOffset()` methods |
| 23 | + |
| 24 | +**Batch Operations**: |
| 25 | +- `ZerobusStream.ingestRecords(Iterable)` - Ingest multiple records with single acknowledgment |
| 26 | +- `ZerobusStream.getUnackedBatches()` - Get unacknowledged records preserving batch grouping |
| 27 | +- `EncodedBatch` class - Represents a batch of encoded records |
| 28 | + |
| 29 | +**Arrow Flight Support** (Experimental): |
| 30 | +- `ZerobusArrowStream` class - High-performance columnar data ingestion |
| 31 | +- `ArrowTableProperties` class - Table configuration with Arrow schema |
| 32 | +- `ArrowStreamConfigurationOptions` class - Arrow stream configuration |
| 33 | +- `ZerobusSdk.createArrowStream()` - Create Arrow Flight streams |
| 34 | +- `ZerobusSdk.recreateArrowStream()` - Recover failed Arrow streams |
| 35 | + |
| 36 | +**New Callback Interface**: |
| 37 | +- `AckCallback` interface with `onAck(long offsetId)` and `onError(long offsetId, String message)` |
| 38 | +- More detailed error information than the deprecated Consumer-based callback |
| 39 | + |
| 40 | +### Deprecated |
| 41 | + |
| 42 | +- `ZerobusStream.ingestRecord(RecordType)` - Use `ingestRecordOffset()` instead. The offset-based API avoids CompletableFuture allocation overhead. |
| 43 | +- `ZerobusStream.ingestRecord(IngestableRecord)` - Use `ingestRecordOffset()` instead. |
| 44 | +- `ZerobusStream.ingestRecords(Iterable)` - Use `ingestRecordsOffset()` instead. |
| 45 | +- `ZerobusStream.getState()` - Stream state is no longer exposed by the native backend. Returns `OPENED` or `CLOSED` only. |
| 46 | +- `ZerobusStream.getUnackedRecords()` - Returns empty iterator. Use `getUnackedBatches()` or `getUnackedRecordsRaw()` instead. |
| 47 | +- `StreamConfigurationOptions.Builder.setAckCallback(Consumer<IngestRecordResponse>)` - Use `setAckCallback(AckCallback)` instead. |
| 48 | +- `ZerobusSdk.setStubFactory()` - gRPC stub factory is no longer used with native backend. Throws `UnsupportedOperationException`. |
| 49 | + |
| 50 | +### Platform Support |
| 51 | + |
| 52 | +- Linux x86_64: Supported |
| 53 | +- Windows x86_64: Supported |
| 54 | +- macOS: Not yet supported (planned for future release) |
| 55 | + |
3 | 56 | ## Release v0.1.0 |
4 | 57 |
|
5 | 58 | Initial release of the Databricks Zerobus Ingest SDK for Java. |
|
0 commit comments