v4.1.0

Added

• fastproto bundle module: New fastproto-bundle module (artifactId: fastproto) that aggregates core and processor, allowing users to include just one dependency for full functionality
• Integration tests for the bundle module to verify aggregated functionality

Removed

• Removed CodecProcessor and @GenerateCodec annotation (limited practical value)
• Removed runtime dynamic compilation (FormulaBuilder and related classes); lambda formulas now rely entirely on compile-time code generation

Changed

• Project restructured into three modules:
  • fastproto-core - Core library
  • fastproto-processor - Annotation processor
  • fastproto (bundle) - Aggregator module, recommended for users
• Updated documentation to simplify dependency references

Migration Guide

Users upgrading from 4.0.0 now only need one dependency:

Maven:

<dependency>
    <groupId>org.indunet</groupId>
    <artifactId>fastproto</artifactId>
    <version>4.1.0</version>
</dependency>

v4.0.0

Breaking Changes

• Multi-module restructure: Project split into two modules:
  • fastproto - Core library with all runtime functionality (backwards compatible artifactId)
  • fastproto-processor - Annotation processor for compile-time formula generation
• Dynamic lambda compilation is now disabled by default for better Android/JRE compatibility

Added

• fastproto-processor module with SPI auto-discovery for annotation processors
• FormulaProcessor - Generates Function classes at compile time for @DecodingFormula(lambda="...") and @EncodingFormula(lambda="...")
• CodecProcessor - Generates codec wrapper classes for @GenerateCodec annotated classes
• FormulaRegistry - Runtime lookup for compile-time generated formula classes
• Full Android and Java 11+ JRE compatibility for lambda formulas via annotation processor

Changed

• Lambda formulas now use compile-time code generation instead of runtime dynamic compilation
• Users need to add fastproto-processor as a dependency to use lambda formula features
• Updated documentation for new multi-module structure

Migration Guide

For users upgrading from 3.x:

v3.12.3

Added

• New @BcdType annotation and BcdCodec for fixed-length packed BCD integers (with byte order support), mapped to int / Integer.
• Unit tests and FastProto integration tests for BCD, covering big-endian / little-endian and error conditions.
• New @Expect / @Expects annotations for constant assertions at fixed offsets; verify on decode and auto-fill constants on encode.

Changed

• README and Quick Start: added side-by-side examples comparing manual bit/offset parsing to FastProto’s annotation-based approach, plus an explicit “Design Philosophy” section.
• Docs site (docs/index.html, docs/annotation-mapping.md, etc.): documented BCD type mapping and behavior, and refreshed core feature descriptions.
• Core resolving pipeline: extended dynamic lengthRef support so computed lengths are bound to preceding fields and propagated through Resolver / CodecFlow into string/array/struct-array codecs.
• Core resolving pipeline: extended dynamic offsetRef support so offsets can reference numeric fields; Resolver binds offset suppliers and CodecFlow wraps annotations, making offset honor runtime values during encode/decode.

v3.12.2

Added

• @AutoType now supports lengthRef for variable-length strings/arrays/collections with auto type inference.
• Checksum enhancements: range-based checksum calculation support.
• Documentation: added docs/variable-length.md (variable length & struct arrays) and new integration pages docs/netty-integration.md, docs/kafka-integration.md.
• Variable-length support: via lengthRef referencing a count field, enabling variable-length strings/arrays and struct arrays.

Changed

• Proxy forwarding: ProtocolType.proxy(AutoType, ...) forwards lengthRef to target annotations (e.g., StringType, *ArrayType, StructArrayType).
• Docs & examples: help sidebar includes “Variable Length & Struct Arrays”; installation snippets updated to 3.12.2.

Fixed

• DateCodecTest stability: truncate Instant/LocalDateTime to milliseconds.

v3.12.1

Fixed

• ByteBuffer: auto-growth when writing beyond current capacity on non-fixed buffers to prevent ArrayIndexOutOfBounds.
• ByteBufferIOStream.align: validation now requires a positive power-of-two alignment; incorrect values throw IllegalArgumentException.

Changed

• Docs site: redesigned docs/index.html homepage (overview, features, basic usage), set Quick Start as default in help, added icons to features.
• Version references in docs updated to 3.12.1.

Added

• Android compatibility guide docs/android.md and cross-links.

v3.12.0

• Single-annotation checksum

  • New @Checksum(start, length, offset, type, byteOrder) defines the compute range and storage location in one place.
  • Works seamlessly in both encode (auto-write) and decode (auto-verify) flows.

• Broader algorithm coverage

  • Added: CRC8_SMBUS, CRC8_MAXIM, XOR8 (BCC), LRC8, CRC16_MODBUS, CRC16_CCITT, CRC32C, CRC64_ECMA-182, CRC64/ISO.
  • Endianness for stored values honored via byteOrder; computation range is endianness-agnostic.

• Performance & robustness

  • ChecksumUtils reuses singleton engines to avoid per-call allocations.
  • Explicit error on unsupported type (CodecException).

v3.11.0

1. Introduced checksum annotations and pipeline support (ChecksumFlow, Pipeline improvements).
2. Added network and IoT usage examples.
3. Implemented double-read tests for ByteBufferInputStream.
4. Add checksum.

v3.10.3

Fix bugs when using @DecodingIgnore & @EncodingIngore.

v3.1.3

Fix bugs when using @DecodingIgnore & @EncodingIngore.

v3.10.2

1. remove CodecUtils;
2. add DecodeUtils and EncodeUtils;
3. update README.md