Types, TypeScript and the JIT

on YouTube ↗10m

" summary

Concrete claim: TypeScript never hurts JIT performance and can make type behavior more predictable but offers no guarantees because TypeScript types don't always match engine representations (e.g. TS "number" vs engine 32‑bit ints and doubles and an array-out-of-bounds can still be undefined). Implementation details: spider monkey uses inline caches and native stubs, two-tier JIT (Baseline and Ion compilers), and bails out to the interpreter when assumptions fail, sometimes throwing away compiled code; inline caching yields about 2× speedup in common workloads.

AI-written summary · about

" chapters (7)

1. 0:00Introduction
2. 1:12Russell’s Paradox and types
3. 2:35Type checking and inference
4. 3:38TypeScript design goals
5. 5:04JIT compilation fundamentals
6. 6:48Dynamic dispatch and caching
7. 8:55TypeScript and JIT performance

" related talks

• 26mJavaScript JITsDavid Anderson, David Mandelin · JSConf EU 2011
• 18mDemystifying JS EnginesAlejandro Oviedo · CascadiaFest 2016
• 17mDemystifying (JavaScript) enginesAlejandro Oviedo García · JSConf Iceland 2016
• 24mTypeScript and how I've learned to love the compilerFelix Rieseberg · Shift Dev 2019
• 01h 02mExploring TypeScriptVenkat Subramaniam · Great Indian Developer Summit 2015
• 40mTypeScript: State of the UnionJames Henry · Web Unleashed 2018
• 57mTypeScript : More than just another JS transpilerBasarat Ali Syed · NDC Sydney 2016
• 39mTypeScript: Why? When? How?Félix Billeau, Sylvain Pontereau · DevFest Paris 2019