Wasm vs JavaScript benchmark Browse Catalog Experiment Evidence data v1

Text & Parsing · CMS Pipeline · text.markdown-cms.v1

Markdown CMS Render Pipeline

Parses Markdown and raw HTML subsets, builds table-of-contents anchors, enforces security allowlists, and serializes escaped canonical HTML.

Run Benchmark Loop

Ready. Select target and loop iterations, then click Start.

Correctness evidence

No performance claim. This bounded demo runs one target in a worker. It does not upload or store the source or result. Canonical HTML is shown as text and is never inserted into the page.

Controlled target

Source accepts at most 40,960 UTF-8 bytes and 4,096 non-empty lines. Raw HTML permits only attribute-free em and strong elements without nested markup. Links permit only https://example.test/ and https://docs.example.test/; figure sources permit only https://images.example.test/. The worker stops after 10 seconds.

Text result

Ready.

No result yet.

What it does

The fixed grammar parses headings, paragraphs, raw HTML, whole-line links, and figures. It then creates table-of-contents anchors, checks the URL and element allowlists, and serializes escaped canonical HTML. The Wasm option performs those stages in authored linear-memory code.

Why compare Wasm vs JS

Text parsing, character scanning, and string building.