How darnstrom/daqp works
Project type and purpose
• DAQP is an open-source, library-free C solver for convex (and certain mixed-integer) quadratic programs.
• It exposes a pure-C core engine (the solver) and multi-language wrappers (Julia, MATLAB, Python, Eigen/C++, Simulink).
File structure
• Top level:
• CMakeLists.txt + include/ + src/ → the core solver library (API headers in include/, implementation in src/).
• codegen/ → a small utility (codegen.c/h) to auto-generate code (likely builds custom solver instances).
• interfaces/ → subdirectories for each language interface:
– daqp-eigen (C++ wrapper using Eigen)
– daqp-julia
– daqp-matlab
– daqp-python (Cython)
– daqp-simulink
• docs/ → static site (Jekyll/GitHub Pages) with installation and usage guides.
• .github/workflows/ → CI pipelines for building, testing, packaging wheels, releases, etc.
README for additional insights
• Core functionality: solve quadratic programs with bound or affine constraints.
• Interfaces: high-level language bindings rely on the core C library.
• CI badges and docs link indicate automated testing across languages and platforms.
Final advice
• Be very detailed and capture every distinct module.
• Lay out the core at center, cluster related wrappers around it, position CI/CD and docs at edges.
• Special build steps (CMake configuration, pyproject.toml, Project.toml, MEX compilation).
• The goal is maximal clarity on how the multi-language components all depend on the single C solver core.