v2.1.0: Everything a document is made of

A release about fidelity rather than reach. What a document is made of — diagrams, formulas, code, the pictures — used to be the part that did not survive the trip.

In this release

• Mermaid diagrams are drawn rather than printed as code — in the preview, the downloaded .html and in print, redrawn when the theme changes
• Maths between dollar signs, and LaTeX's own \( \) and \[ \], typeset with KaTeX as MathML — so a formula needs no stylesheet and no fonts
• Code is highlighted in every fence that names a language, in the document's own palette, inside the exported file
• The things people actually write in Markdown — front matter, GitHub alerts and Obsidian callouts, footnotes, wikilinks, ==highlight==, H~2~O
• PowerPoint → Markdown, the eleventh conversion: a slide is a section, in the order the deck plays, and the speaker notes come with it
• Pictures survive an export — images inside a Notion, Confluence, Obsidian or PowerPoint archive are carried into the document, and Word's are kept instead of thrown away
• A document says what is wrong with it — dead anchors, duplicate headings, empty headings, pictures with no alt text, each with its line
• The browser extension reads what the screen shows rather than what the markup claims: hidden routes, shadow DOM, typed form values, preformatted blocks kept preformatted, a side panel that follows a route change, and a fail-safe that refuses to hand back an emptied page

For the extension

The manifest reads package.json, so this tag is the package: brand/store/transformpipe-2.1.0.zip for Chrome and transformpipe-firefox-2.1.0.zip for Firefox, both built by npm run ext:zip. BUILD.md reproduces either one from this tag.

194 commits since 2.0.0. Everything that shipped between the two is on the changelog, which lists the releases and the changes between them: https://transformpipe.com/changelog