Knit, Chisel, Hack: Building Programs in Guile Scheme
summary
Guile claims to scale from throwaway scripts to long‑lived systems by combining Scheme with POSIX and a rich workbench: a REPL with debugger, profiler, disassembler, macroexpander and optimizer, plus ELF module output and ~2 MB startup. Benchmarks show allocation ~1 GB/s and instruction retire rates up to ~400–500M/sec; microbenchmarks put Guile dozens of times slower than tuned C but similar order to Python. Built items include a web server (~10,000 requests per second per core), a ping loop (≈50k back‑and‑forths/sec/core) and a library using delimited continuations to implement lightweight fibers for concurrency.
AI-written summary · about
chapters (8)
- 0:00Start
- 1:03crafty pleasures
- 2:01what's
- 11:05from scripts
- 19:56macros extend language syntax
- 24:09prompts
- 31:34then deploy
- 33:10godspeed!
related talks
29mThe thin CLIent approach
43mCompiling like a boss!
59mThe Hitchhikers guide to faster builds
21mFrom Laptop to Lambda: Outsourcing Everyday Jobs to Thousands of Transient Functional Containers
17mConceptualizing Large Software Systems
01h 00m“Practical Performance Practices"
19mJumping on the Bandwagon: How I ended up writing Gleam for a living
23mPolyglot Builds with Bazel