Proprietary extensions: Microsoft Internet Explorer accepts a
proprietary extension to HTML that looks like malformed mark-up:
"<![if xxx]>" or "<![endif]>" (where xxx is text that doesn't contain
"]"). The parser currently reports (correctly) a syntax error. Should
there be a mode in which this is accepted? If so, as what? as TEXT?


Let hxindex recognize DATA-INDEX attributes in addition to TITLE? 
(These new DATA- attributes are suggested by HTML5.)

Let hxindex look at locale for sorting order?

Let hxnormalize sort the <head> into a canonical order? (title, meta
sorted by name and http-equiv, link sorted by rel, style, script)

Add ruby, rb, rp, rt and rtc to dtd.hash.

Add MathML elements to dtd.hash.

An option to add <![CDATA[...]]> to CDATA sections or to escape <, >
and & in such sections (only when output is XML).

Properly parse the srcset attribute.
