Keynote: Memory of Past and Future – Wasm’s Evolving Model of Store
summary
Wasm’s store evolves from a single flat byte memory to a multi-component store: ref types and generalized tables keep opaque pointers out of linear memory, GC structs and arrays give typed heap allocation (and i31 unboxed integers) so engines can reuse built-in collectors, and upcoming features add shared memory, multiple memories per module, 64-bit addressing and custom page sizes plus atomics. The design stresses formal, defined semantics (no UB), gradual typing and compatibility; some features shipped, others paused for security and complexity but prototypes and specs exist.
AI-written summary · about
chapters (10)
- 0:00WebAssembly memory model
- 6:30Why WebAssembly is a low-level virtual machine
- 10:18WebAssembly memory myths
- 17:22Wasm 1.0 memory limits
- 20:56Wasm 2.0 reference types
- 26:57Wasm 3.0 multiple memories
- 32:37Wasm needs garbage collection
- 43:58Compiling object-oriented languages to WebAssembly
- 49:17Wasm threading & memory
- 58:10Wasm 3.0 features & future
related talks
43mRichWasm: Bringing Safe, Fine-Grained, Shared Memory Interoperability to WebAssem...
22m[Keynote] Wasm GC and the Future of High-Level Language Compilation
47m[Keynote] Keynote and SpecTec demo
57mIs Wasm Becoming Garbage? (Keynote)
37mBridging the Architecture Divide: How Hardware Limits Wasm, and How We Can Do Better
40mSmarter Operating Systems Will Use Wasm - The Coming OS Revolution
26mKeynote: The Path to Components
43mMeet Your Future - WebAssembly