Cló: The Algorithms of TeX in Clojure
summary
A Clojure reimplementation of TeX’s core algorithms becomes a modular backend rather than a line-by-line clone: the last three stages and a simplified line breaker are implemented and run either single-threaded or as a parallel dataflow pipeline by swapping the threading macro, yielding measurable parallel speedups though overall execution remains much slower than classic TeX. DVI output uses a thread-local VAR for DVI state after the state-monad proved slow; reading TFM font tables uses java.nio scattering reads and is “blazingly fast.” The design shifts toward composable protocols (list/line/page elements, elasticity, curf) for reuse and education.
AI-written summary · about
chapters (10)
- 0:00Start
- 17:00Interesting Questions
- 18:11Discovery
- 20:59TEX's optimizations
- 28:41Stateful I/O
- 30:49java.nio Scattering Reads
- 32:05Utility
- 32:36Composable
- 37:39Green-screen architecture
- 38:34It's 1982, and ...
related talks
01h 15mEffective Programs - 10 Years of Clojure
34mPredictably Fast Clojure
38mLearning Clojure: Next Steps
38mScientific Clojure, a bird's eye view
31mScientific Computing with Clojure
43mDoing Data Science with Clojure
28mEvery Clojure Talk Ever
27mData and Clojure: Using Clojure for more than software development