Line break.
Enter <b>.
Enter <code>.
Enter <em>.
Enter <h1> title.
Enter <h2> title.
Enter <h3> title.
Enter <h4> title.
Enter <h5> title.
Enter <h6> title.
Enter <i>.
Enter <img>.
Enter <li>.
Enter <ol>.
Enter <p>.
Enter <pre>.
Enter <strong>.
Enter <ul>.
Exit </b>.
Exit </code>.
Exit </em>.
Exit </h1> title.
Exit </h2> title.
Exit </h3> title.
Exit </h4> title.
Exit </h5> title.
Exit </h6> title.
Exit </i>.
Exit </img>.
Exit </li>.
Exit </ol>.
Exit </p>.
Exit </pre>.
Exit </strong>.
Exit </ul>.
You MUST make that call before getting the bytes output of the renderer. No subsequent can be made with that IFlowDocument.
Next page.
Output text.
A Flow Document produces output without any box model, in a streamed manner. If something fits, it is included. Honestly, it's already complicated and having boxes and defering rendering is probably better for better results. For example, this rendere can't ever support hyphenation or text justifying. The interface is thought to be able to render Markdown quickly.