XML5

Background: standards engineer, browsers, W3C & WHATWG, HTML, CSS, APIs (DOM, XMLHttpRequest)…

Objective: Develop a revised version of XML that does away with well-formedness in favor of well-defined error handling.

XML was one of those child prodigies pushed to do all sorts of things. Then realised it had to make a choice. — @jenit

Benefits

Developers have difficulty deploying XML robustly.

I.e., most are “bozos” as Tim Bray would say.

If the web is to migrate to XML, making XML closer to HTML is the only way.

Migration becomes trivial.

Design

Full backwards compatible. Existing XML documents should be identically parsed.

Streaming should be possible.

Not require schemas.

Fully deterministic.

Error handling considerations

Input stream: if in e.g. hz-gb-2312 "~" is followed by "A" what happens is currently undefined.

Tokenizer: unquoted attributes, illegal characters in tags, …

Tree builder: missing end tags, dealing with namespaces, … new features?

Closing thoughts

Thanks! (@annevk)