Lynx

From Visible Text to Semantic Documents

A second document representation preserved headings, lists, emphasis, and other meaning without placing markup in the writing surface.

On 22 May, the repository introduced LynxDocument, a semantic model that sits beside the visible editor text. Titles, headings, paragraphs, quotes, lists, inline styles, math, citations, and template references could now be represented as document meaning rather than punctuation inserted into the page.

The difficult part was not defining the model. It was keeping ordinary editing and structured state aligned. The synchronization strategy became intentionally conservative: accept supported changes within one block, preserve untouched structure, and report an out-of-sync state when a change cannot be reconciled safely.

That tradeoff still shapes Lynx. A dual representation lets the editor feel like normal writing while enabling structured output. It also creates a boundary that the product must explain and continue to harden, especially for edits that cross blocks.

← Back to Lynx