Re: Footnote discussions

How the markup should look is an interesting question. Lots of "classic"
XML/SGML vocabularies have a footnote element that essentially serves as
both marker and reference:

<para>Call me Ishmael.<footnote><para>This is a
footnote</para></footnote></para>

When rendered, the footnote element is moved somewhere, and a reference
left in its place.

The PDF formatters Prince and AntennaHouse can do something similar today
with HTML footnotes:

<p>Call me Ishmael.<span class="footnote">This is a footnote</span></p>

They use CSS to move the footnote and generate the references using CSS
counters and some non-standard pseudo-elements:

span.footnote { float: footnote }
span.footnote::footnote-call { content: counter(footnote); vertical-align:
super; }
span.footnote::footnote-marker { content: counter(footnote) '. '; }

This leads to content model problems, as often footnotes may have
sectioning or block elements inside them (like blockquote, div, etc.) which
are not valid inside paragraphs.

We've also run into problems with implementing footnotes using CSS regions.
It's easy enough to move the footnote element somewhere else, but CSS has
not defined ways of leaving markers behind to serve as the callout.

Separating the note and the reference has its own issues, as you have to
essentially create all the links yourself, and it seems a bit arbitrary
where to put the note content itself.

I'd lean towards having a note element that could be embedded at the point
of reference, with a rather expansive content model. A noteref element
would be optional, for the case where there are multiple references to the
same note. But CSS would need to address the issue of leaving a reference
when the note element was moved to somewhere else in the document. Such a
pseudo-element (one possible proposal) would be more generally useful, with
sidebars, images, and other such content.

Dave