Translating research algorithms into production systems : Adding sorting to EmbedDB

Embedded systems with constrained random access memory require external sorting algorithms that carefully manage memory across buffer allocation and SD file read/write operations. This thesis advances the EmbedDB sorting implementation to a production-ready state by introducing improved interfaces, robust memory management, and comprehensive testing. The sorting features also allow the implementation of GROUP BY on any column that can be used in conjunction with the existing aggregate methods and queries. Benchmarks between EmbedDB and SQLite sorting on different devices is performed, showing faster speeds for SQLite but improved relative performance for EmbedDB on less performant hardware.