Google Open-Sources FlatBuffers: Efficient, Cross-Platform, Serialization Library
The Google "Fun Propulsion Labs" team has recently open-sourced FlatBuffers. Built especially to support performance needs of game developers, FlatBuffers stores serialized data in buffers which can be either stored in files or transferred across the network as-is, without any parsing overhead.
Some key features of FlatBuffers -
- Access to serialized data without packing/unpacking
- Memory efficiency and speed - only memory required to access data is that of the buffer; 0 additional memory required.
- Flexibility - forwards and backwards compatibility via optional fields
- Tiny Code footprint
- Strongly Typed - errors caught at compile-time instead of run-time
- Convenience - Terse code via generated C++ headers. Optional functionality to parse schemas and JSON-like representations at runtime efficiently if needed (faster and more memory efficient than other JSON parsers)
- Cross-Platform: Written in C++ with no dependency outside of STL, so can be used on any platform that has a C++ compiler. Currently the project includes methods to build and examples to use the library for Android, Linux, Windows and OSX.
The advantage of FlatBuffers over alternative solutions such as Protocol buffers or JSON parsing, is that there is less overhead, especially due to the lack of parsing.
Using FlatBuffers in C++ involves following steps -
- Create schema file using the specified DSL
- Use the schema compiler to generate C++ header files
- Include the generated header file in your program (also needs the flatbuffers/flatbuffers.h file to be in your path)
- Read and Write data using the FlatBufferBuilder class (which constructs a flat binary buffer)
There are still a few rough edges, as evident by early discussions on the discussion group. Support for Java is also experimental at this stage. C# support is one of the most popular requests, and might get supported in the next release.
You can learn more about FlatBuffers by referring the documentation.
Another project that avoids encoding/decoding step to achieve performance benefits and is worth a look is Cap'n Proto.
A lot of serialize format!