The non-translatable features are extremely niche, though. To continue the quote: > For example, namespaces cannot be represented using the HTML syntax, but th

The non-translatable features are extremely niche, though. To continue the quote: > For example, namespaces cannot be represented using the HTML syntax, but they are supported in the DOM and in the XML syntax. Similarly, documents that use the noscript feature can be represented using the HTML syntax, but cannot be represented with the DOM or in the XML syntax. Comments that contain the string "-->" can only be represented in the DOM, not in the HTML and XML syntaxes. I.e. the big thing you'd fail to capture when doing an automated HTML -> parsed DOM -> XML trip is noscript tags. No big loss. Others I can remember are similarly tiny, e.g. some issues around textarea elements containing only newlines as their initial contents. Some discussion at https://html.spec.whatwg.org/multipage/parsing.html#parsing https://html.spec.whatwg.org/multipage/parsing.html#parsing Ctrl+F "roundtrip", although note that most of that is actually about what happens if you construct weird DOMs with JS and try to do a DOM -> serialized HTML -> reparsed DOM roundtrip. I think it's pretty safe to say in general that the procedure of "parse HTML to DOM, serialize to XML" will preserve everything interesting about a document. Especially if that document is already valid HTML.