alipay/fury

Features

• Binary Deserialization - Reconstructs original typed objects from binary payloads using a shared schema to ensure type safety.
• Language-Neutral Data Serialization - Implements a portable binary wire format for encoding data structures independently of specific languages or architectures.
• Cross-Language Serialization Frameworks - Implements a high-performance multi-language binary serialization framework for encoding domain objects and complex graphs.
• Interface Definition Languages - Provides an IDL compiler to generate type-safe native models and serialization boilerplate across multiple languages.
• Cross-Language Data Model Definitions - Uses a schema IDL to generate native data structures across multiple languages to ensure a consistent data contract.
• Schema Definition - Uses a language-neutral schema format to specify structured data types, unions, and services.
• Schema-to-Native Type Mappings - Maps native language types to stable, namespaced identifiers for consistent data exchange between services.
• Data Schema Definitions - Provides declarative systems for defining data structures that ensure consistent layout across different programming languages.
• Graph Serializers - Provides a serialization engine that preserves circular references and referential integrity within complex object hierarchies.
• Language-Specific Type Definitions - Generates native programming language types based on schema definitions to ensure consistent field mappings.
• Schema Evolution - Maintains forward and backward compatibility by evolving data structures without requiring destructive migrations.
• Analytics-Optimized Row Formats - Uses a row-oriented binary encoding optimized for high-throughput processing and zero-copy data access.
• Random Field Accesses - Allows reading specific fields from binary records without performing full object deserialization.
• IDL Model Generation - Produces compiler output for data models and schema modules based on an Interface Definition Language.
• Zero-Copy Data Access - Implements zero-copy mechanisms to access serialized data structures directly from memory without full deserialization.
• Zero-Copy Deserialization - Ships a zero-copy binary reader that allows accessing specific fields without deserializing the entire object.
• Native AOT Compilation - Generates optimized serializer code at build time to eliminate runtime reflection and reduce memory usage.
• Serialization Code Generators - Generates static serialization logic at build time to eliminate runtime reflection and improve performance.
• Cross-Language Data Mapping - Provides bidirectional transformation of complex data structures between different programming languages.
• Cross-Platform Language Bindings Generators - Transforms schema definitions into type-safe native language bindings for multiple target platforms.
• Object Serialization - Encodes complex object graphs with circular references and polymorphic types while maintaining structural integrity.
• IDL Compilers - Translates interface definition language files into native source code for multiple target languages.
• Compact Binary Object Encoding - Converts domain objects into a compact binary format using variable-length encoding to reduce payload size.
• Reference Tracking - Tracks object identities during serialization to preserve shared references and resolve circular dependencies.
• Polymorphic Type Mapping - Implements automatic type dispatching to preserve polymorphic object types across the binary wire format.
• Serialization - Preserves inheritance hierarchies by including base class fields within derived types during binary serialization.
• Schema Definition Languages - Includes a compiler that translates schema definitions into native types and serialization boilerplate across multiple languages.
• Multi-Language Model Generators - Produces type-safe data models and serializers for a variety of target programming languages from shared schemas.
• Primitive Array Encoding - Writes dense numeric or boolean data as raw binary buffers to achieve high performance throughput.
• Apache Arrow Processing - Converts serialized row-based data into Apache Arrow columnar formats to enable high-performance analytical workloads.
• Binary Encoding Specifications - Assigns specific binary formats to scalar types using language-specific annotations and metadata.
• Language-Specific Optimizations - Provides high-performance serialization alternatives for complex native types, including support for lambdas.
• Row-to-Columnar Conversions - Transforms row-based binary data into Apache Arrow RecordBatch formats for high-performance analytical workloads.
• Collection Serialization - Encodes grouped data structures like lists and maps using size markers, type metadata, and reference tracking.
• Deep Object Manipulations - Provides utilities for performing recursive deep clones of complex object graphs while preserving referential integrity.
• Zero-Copy - Utilizes zero-copy buffer paths for large binary payloads to eliminate unnecessary memory copying.
• Field-Level Metadata Annotations - Attaches metadata to fields using annotations to provide encoding hints and reference tracking.
• gRPC Service Generators - Automatically generates gRPC service code and boilerplate from protocol definitions during the build process.
• Metadata Deduplication - Deduplicates field names, type names, and namespaces within a session to reduce binary payload size.
• gRPC Implementations - Provides a high-performance gRPC protocol implementation for remote procedure calls.
• Out-of-Band Buffer Referencing - Optimizes binary data transport by referencing external buffers instead of inlining large payloads.
• Serializer Registrations - Allows explicit mapping of classes to specialized serializer implementations for optimized or non-standard encoding.
• Class Registration Optimizations - Optimizes binary payload size by mapping classes to numeric identifiers with fallback options.
• Whitelisted Deserialization - Prevents security exploits by restricting object instantiation during deserialization to a registry of trusted types.
• Message Type Registrations - Assigns unique identifiers to user-defined structures to enable precise identification during deserialization.
• Deserialization Security - Controls class instantiation during decoding through explicit registration and configurable security policies.
• Serialization Performance Tuning - Increases throughput using JIT code generation, SIMD acceleration, and variable-length encoding.
• Singleton Identity Preservation - Ensures that singleton objects maintain referential equality after being deserialized from a binary format.
• Dynamic Type Serialization - Processes generic object payloads without requiring a predefined static type at compile time.
• Runtime-Specific Serialization Paths - Uses native serialization paths tailored for specific language runtimes to achieve faster processing.
• Stream Metadata Deduplication - Reduces redundant type information by writing class metadata once per stream and using local index references.
• Unknown Type Preservation - Preserves data for classes that do not exist in the local runtime to allow lossless round-tripping of payloads.
• Serialization - Blazing fast JIT-powered zero-copy serialization framework.

Star history