flux-md 0.3.1 → 0.3.2

package/CHANGELOG.md

@@ -4,6 +4,16 @@ Notable changes to flux-md. Format based on
 [Keep a Changelog](https://keepachangelog.com/); this project aims to follow
 [Semantic Versioning](https://semver.org/).
 
+## 0.3.2 — 2026-05-27
+
+### Documentation
+
+- Rewrote the README to describe flux-md on its own terms and removed all
+  references to and comparisons with other libraries. No code changes — the
+  published API and behavior are identical to 0.3.1.
+- Fixed the React quick-start example: import `useEffect` and guard the async
+  append loop so it can't run after unmount or a stream change.
+
 ## 0.3.1 — 2026-05-27
 
 ### Performance
@@ -34,12 +44,11 @@ Notable changes to flux-md. Format based on
   `<div class="math math-display">`) carrying the LaTeX as text content — bring
   your own KaTeX (flux-md stays zero-dep) or override `components.MathBlock`
   (which receives the LaTeX as `text`). Display fences are blank-line tolerant
-  and stream incrementally. Addresses [Streamdown #522]. Off by default.
+  and stream incrementally. Off by default.
 - **`dirAuto`** — opt-in per-block `dir="auto"` on block-level text elements
   (`p`, `h1`–`h6`, `blockquote`, `ul`/`ol`/`li`, `table`, alerts, footnotes), so
   the browser detects each block's direction (RTL/LTR) independently in
-  mixed-language documents. Code blocks stay LTR. Addresses [Streamdown #509].
-  Off by default.
+  mixed-language documents. Code blocks stay LTR. Off by default.
 
 ### Performance
 
@@ -67,6 +76,3 @@ Notable changes to flux-md. Format based on
 - Initial public release: zero-dep streaming markdown, Rust→WASM core, one Web
   Worker per stream, CommonMark 0.31 (652/652) + GFM (tables, strikethrough,
   task lists, extended autolinks, GitHub alerts, footnotes).
-
-[Streamdown #522]: https://github.com/vercel/streamdown/issues/522
-[Streamdown #509]: https://github.com/vercel/streamdown/issues/509

package/README.md

@@ -2,7 +2,7 @@
 
 Zero-dep streaming markdown for the browser. Rust→WASM core, one Web Worker per stream, incremental parse with speculative closure for mid-stream constructs.
 
-Built because [Streamdown](https://streamdown.ai) crashes the main thread when you run 5 concurrent LLM calls. Same input, this library uses **~8× less peak heap, ~6× less retained memory, and ~2× less main-thread blocking** — measured in-browser with `performance.memory`. See [the live demo](https://md.hsingh.app/) for an A/B comparison.
+Parsing runs entirely **off the main thread** — each stream gets its own pooled Web Worker, so many concurrent LLM responses render without contending for the UI thread. On each token the parser re-parses only the **active tail**, not the whole document, and heavy renderers (syntax highlighting, math, mermaid) are **deferred until a block closes**. The result is low retained memory and a main thread that stays responsive while streaming. See [the live demo](https://md.hsingh.app/).
 
 ## Install
 
@@ -37,18 +37,27 @@ client.finalize();
 In React:
 
 ```tsx
-import { useMemo } from "react";
+import { useEffect, useMemo } from "react";
 import { FluxClient, FluxMarkdown } from "flux-md";
 
 export function ChatMessage({ stream }: { stream: AsyncIterable<string> }) {
   const client = useMemo(() => new FluxClient(), []);
+
   useEffect(() => {
+    let cancelled = false;
     (async () => {
-      for await (const chunk of stream) client.append(chunk);
-      client.finalize();
+      for await (const chunk of stream) {
+        if (cancelled) return; // stream changed / unmounted mid-flight
+        client.append(chunk);
+      }
+      if (!cancelled) client.finalize();
     })();
-    return () => client.destroy();
+    return () => {
+      cancelled = true;
+      client.destroy();
+    };
   }, [stream]);
+
   return <FluxMarkdown client={client} />;
 }
 ```
@@ -57,7 +66,7 @@ Multiple concurrent streams just need multiple clients — each runs in its own
 
 ## What it does
 
-| Concern | flux-md | typical react-markdown / Streamdown |
+| Concern | flux-md | conventional main-thread renderer |
 |---|---|---|
 | Re-parse on each token | No — only the active tail | Yes, full string |
 | Where parse runs | Web Worker (off main thread) | Main thread |

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "flux-md",
-  "version": "0.3.1",
+  "version": "0.3.2",
   "description": "Zero-dep streaming markdown for the browser. Rust→WASM core, Web Worker per stream, incremental parse with speculative closure.",
   "type": "module",
   "main": "./src/index.ts",

package/src/hi.ts

@@ -5,8 +5,8 @@
  * languages fall through to plain escaped text. ~6KB minified.
  *
  * Highlighting is per-block, runs once when the block closes. We never
- * highlight an open (streaming) block — that's what gives us the big perf
- * win vs Streamdown's per-chunk Shiki invocation.
+ * highlight an open (streaming) block, which avoids re-highlighting the same
+ * code on every chunk — the main perf win for streaming code.
  */
 
 const KEYWORDS_JS = new Set(