How to Flatpack Programs
summary
Argues for treating packaging as product and shows concrete wins: a browser implementation at 10% of size (Ladybird), a terminal re‑write that becomes 100x faster then another tweak making it 10× faster again, and language/runtime projects (Zig, Rock) that land at hundreds of thousands or tens of thousands of lines; examples of full systems and OSes in ~20,000 lines (dusko, Olm/Oma experiments); points to SQLite and tiny libraries as practical tools, measures exabytes of traffic and carbon as the scale of the waste.
AI-written summary · about
chapters (12)
- 0:00Introduction
- 2:06Nature’s packing inefficiencies
- 4:36Flatpack furniture and shipping
- 7:52Software bloat defined
- 12:28Why software bloats
- 15:19Browsers as OS alternatives
- 18:32Dependency icebergs
- 22:05Small code primitives
- 26:42Hardware truth and performance
- 29:57Databases and SQLite
- 33:02Small software in industry
- 34:34Q&A
related talks
01h 02mThe Wrong Kind of Fast: Embracing Evidence-Based Architecture
26mTheory of programming: packaging principles
28mLa nouvelle vague
27mFacing Frontend's Existential Crisis
01h 28mApps at Scale: An Industry Discussion
58mIntentional Code - Minimalism in a World of Dogmatic Design
48mTurbopack: DX Matters
01h 01mLessons Learned from Packaging 10,000+ C++ Projects
