This site

Press / on any page, or the search button at the top, and type. Every word of every page is searchable, including titles, descriptions and code blocks. A result shows the page, the heading the match sits under, and the text around it. Opening it takes you straight to that heading.

What counts as a match

Every word you type counts, even common ones like “the”. Each word also matches its plural and its -ed and -ing forms. The last word matches any word it begins, so results narrow as you type.

A word that appears nowhere on the site is treated as a typo. It matches words one letter away if it’s four to seven letters long, and two letters away if it’s longer. Pages that hold more of your words rank higher, and a word in a page’s title counts extra.

How it runs

Nothing leaves your browser. When the site is built, every page’s text goes into one file, search.xml. Once a page has finished loading, it fetches that file along with the search engine, a 152 KB WebAssembly module. The engine indexes the whole site in about 20 milliseconds, so by the time you press / there’s nothing left to wait on.

The engine is wasm-static-search. Its ranking code is written in Rust and compiled to WebAssembly.