The libSQL project has attempted to create a fork of the SQLite database focused on being open to community developer participation and promoting innovation beyond the original intent of SQLite. The reason given for the fork is SQLite's rather rigid policy regarding accepting third-party code from the community when there is a need to promote improvements. The code for the fork is distributed under the MIT license (SQLite is in the public domain).
The creators of the fork intend to maintain compatibility with the main SQLite and maintain the same level of quality, keeping the set of test cases and gradually expanding it as new features are added. To develop new functionality, it is proposed to provide the ability to use the Rust language, while maintaining the basic part in the C language. In the event of a change in the policy of the main SQLite project regarding the acceptance of changes, the libSQL developers intend to transfer the accumulated changes to the main project and join its development.
Among the ideas for a possible extension of the functionality of SQLite are mentioned:
- Integration of tools for building distributed databases that work at the level of the library itself, and not through the replication of changes in the file system (LiteFS), and without the development of a separate product (dqlite, rqlite, ChiselStore).
- Optimization for using asynchronous APIs, such as the io_uring interface provided by the Linux kernel.
- The ability to use SQLite in the Linux kernel, similar to the support in the kernel of the eBPF virtual machine, for situations where you need to save data sets from the kernel that do not fit in RAM.
- Support for user-defined functions written in any programming language and compiled into WebAssembly intermediate code.
Source: opennet.ru