How fschaper/palo works
Project type and purpose
• This is a standalone C++‐based MOLAP server (“PALO 5.1”) that exposes a rich API over HTTP/HTTPS for multidimensional OLAP operations (cubes, dimensions, cells, rules, etc.).
• It is not a two-tier “front end + back end” web app but a monolithic server offering a programmable API and tool interfaces (e.g., PaloLoader, HTTP interface, AutosaveTimer).
File structure
• Top‐level directories:
• Api – API definition files and HTML/CSS/JS templates used to generate documentation and endpoint stubs
• Library – all core C++ modules, broken down into subpackages (Collections, Engine, Olap, HttpServer, HttpsServer, Parser, Scheduler, Worker, VirtualMachine, etc.)
• Programs – entry points (palo.cpp, palorun.cpp, PaloLoader, HTTP interface adapters, service wrapper)
• Config, Modules, Resource, CMakeLists.txt, Dockerfile, README, LICENSE, CHANGELOG
• Architectural patterns evident:
• Layered monolith: HTTP layer → Dispatch/Jobs → Worker/Engine layer → Storage/Cache layer
• Job queue / thread‐pool scheduling (Scheduler, Thread)
• Parser/Bytecode generation for rules (Parser, VirtualMachine)
• Separation of HTTP vs. HTTPS servers
README for additional insights
• Build system: CMake, supports Windows (VS2010 SP1) and Linux (gcc/g++, Boost, OpenSSL, ICU, Zlib, gperftools)
• The server executable (palo) binds to a port, serves requests, and persists data on disk