How definitelynotsalesman/hydradb works
Project type and purpose
• HydraDB is a standalone, monolithic C++ relational database engine.
• Its primary responsibilities are fast reads/writes on large datasets (via B+ trees), durability (via Write Ahead Logging), and an eventual goal of supporting concurrent transactions and in-memory caching.
File structure
• Top-level:
• database/ – the storage‐engine sources
– BTree.cpp/.hpp → core B+ tree index implementation
– InputBuffer.cpp/.hpp → command/SQL input reader
– helper/Helper.cpp/.hpp → utility routines (e.g., serialization, logging helpers)
• src/ – entry points
– main.cpp → orchestrates startup, command loop
– installer.cpp → sets up initial on-disk structures (e.g., data files, WAL files)
• Absence of frontend/web/UI code implies a CLI tool or embeddable library.
• No external service or network layer is present—this is a pure storage engine.
README for additional insights
• Clearly defines the core features (B+ Tree, WAL, Pages) and future plans (Lock Manager, Cache).
• Break the system into: CLI, Query Processor, Storage Engine (and its submodules), Persistence Layer, Installer
• Concurrency or durability protocols (e.g., WAL write → fsync → page update)
• Keep it modular so it can expand when you add caching or locks