wasm-livive-net

WASM

https://wasm.livive.net/

Preliminary Plan

An overview of what WebAssembly (WASM) is, plus links to the projects in the livive.net portfolio that use it, with context on why each chose WASM over Node.js or other in-browser options.

Status: draft. Items marked (verify) are inferred from portfolio metadata and need confirming against each repo.

Goals

1. Explain WASM clearly: what it is, what it isn’t, and when to use it.
2. Link each WASM-using project with a “Why WASM here” explanation.
3. Compare WASM against alternatives (Node.js/server, plain JS, WebGL/WebGPU, Web Workers).

Approach

• Static site (HTML, CSS, ES modules), no build step, hosted on GitHub Pages (CNAME already present).
• Style: TBD (match the dark card style of beta.livive.net, or something different).
• Optional: a small live demo (hand-written WASM module vs JS timing comparison).

Site Sections

1. Overview. Portable binary format, sandboxed, near-native speed, compile target for C/C++/Rust/Python and others. Limits: no direct DOM access, needs JS glue, larger downloads, slower cold start.
2. WASM vs. alternatives. Node.js is a server runtime, so the real question is client versus server. Node can also run WASM. Also compared: plain JS, WebGL/WebGPU, Web Workers.
3. Project gallery. Cards with live link, repo link, WASM approach and rationale.
4. (Optional) Live demo.

Projects and WASM Approaches

| Project | WASM route | Why WASM (vs. Node.js / alternatives) | |—|—|—| | Age of Mycology (aom) | Pygbag (Python to WASM) | Keeps the Python game code, no JS rewrite, no server | | EuchreGame | Pyodide (+ WebLLM) | Same deterministic Python engine in browser and tests, no backend latency or cost | | Lokarta (lk.livive.net) | Pyodide (verify) | Reuse Python logic client-side | | LunarLander Policy Lab (rl-livive-net) | Pyodide + ONNX Runtime Web | Python physics plus local RL policy inference, no porting to JS | | P2P Tic-Tac-Toe (p2p-ttt) | sql.js (SQLite to WASM) + OPFS | Real SQL with no backend; IndexedDB has no query engine | | Northstar RAG POC (rag-poc) | Transformers.js (ONNX on WASM) (verify) | Private, on-device RAG with no API cost | | ai_chat_distil_01 | LiteRT.js (TFLite) | On-device small model, no inference server |

Recurring Themes

• Reuse existing Python/C code unchanged.
• Zero backend: free static hosting, no ops, data stays on device.
• Determinism and portability across browser and tests.
• Trade-offs: download size (e.g. Pyodide ~10 MB), cold start, WebGPU often faster for heavy ML.

Milestones

1. Verify each project’s WASM usage against its repo.
2. Draft the content (overview, comparison, per-project blurbs).
3. Build the static site: layout, sections, project cards.
4. Add the optional live demo.
5. Deploy to GitHub Pages and cross-link from beta.livive.net.

Open Decisions

• Visual style.
• Include the live demo or not.
• Content source: hand-written, or generated from the portfolio’s JSON.