How everythingelsewasalreadytaken/fabulous works
Project type and purpose
• Project type: A Python-based toolchain (library + CLI + API) for generating custom embedded FPGA fabrics.
• Purpose:
• Fabric architectures via CSV templates and Python objects.
• Generate HDL (Verilog/VHDL) models of those fabrics.
• Invoke external CAD tools (YosysHQ, nextpnr) for bitstream generation and post-fabrication test.
• Provide a user-friendly CLI shell and a Python API for integration or scripting.
File structure
• Top-level:
• FABulous/ — core source code
• docs/ — documentation sources
• tests/ — unit and integration tests
• fabric_files/ — HDL/VHDL project templates
• .github/ — CI checks, container preparation
• Dockerfile, pyproject.toml, requirements.txt
• Inside FABulous/:
• FABulous_CLI/ — command-line interface modules (shell, synthesis commands, helpers)
• FABulous_API.py — high-level Python API entry points
• FABulous.py — core engine orchestration
• fabric_definition/ — data classes for Bel, Tile, SuperTile, Wire, Fabric, etc.
• geometry_generator/ — compute placement & routing geometry
• fabric_generator/ — parse definitions, generate HDL code (Verilog/VHDL), deal with file creation
• fabric_cad/ — bit-gen and nextpnr model generation (invokes npnr via Python)
• fabric_files/ — stock FPAG templates for user projects in CSV, tcl, Verilog/VHDL
• Configuration & metadata:
• .FABulous/ folder (per project metadata, autogenerated)
• Dockerfile and GitHub Actions YAML for building docs, black formatting, CI testing
README for additional insights
• Front-end shell/API → fabric definition loader → fabric generator → external CAD (Yosys, nextpnr) → test harness
• Emphasizes modularity: templates + custom blocks
• Supports both CLI-driven interactive flow and TCL scripts
Detail is crucial
• .FABulous metadata folder to show where state is kept.