The Nano-parsing Architecture: Sane And Portable Parsing For Perverse Environments
summary
Nano-parsing is a concrete architecture used building the Co-dfns APL compiler that parses millions of lines by turning token streams into a forest of single-element trees and applying many tiny top‑down passes with occasional fixpoint iterations. It decouples error reporting from parsing, supports incremental specialization and rich introspection, eliminates deep recursion, and proves GPU compatible and CPU‑efficient in practice, producing usable error messages and successful whole‑codebase parses.
AI-written summary · about
chapters (5)
- 0:00Introduction
- 4:59Nano-parsing overview
- 9:13Refining parse trees
- 11:39Parser demo
- 26:53Advantages of approach
related talks
31mSmart Nano Stores
59mHow to parse a file
09mA variant of recursive decent parsing
30mConcrete Syntax Metapatterns
41mBuilding Parsers in C#: From Theory to Practice. The Uklon Case
26mOrchestrating AI Agents for Large-Scale Refactoring
54mFractal Architecture
46mIncremental, zero-config Code Nav using stack graphs