HtmlDocletWriter.commentTagsToContent has a visitor that generates content trees that do not correctly support the charCount() method.

This is because DocTree StartElement is translated as

RawHtml: <name
Text: any attributes
RawHtml: >

When counting characters, because the attributes are modeled with a Text node, they are incorrectly included in the count.

It would be better to include all the content in a single well-structured node.