On Fri, 17 Sep 2004, David Woolley wrote:

> > I have understood - and used - it as a row that sums up the data in a
> > table that contains numerical information (such as financial reports)..
>
> It isn't.  It's a running footer for each page of the table, like
> thead is a running header.

That's one opinion. I've seen many opinions, but no authorative definition
of semantics, or even a fair attempt at that.

If the real difference between <thead> and <tfoot> is that the former is
above the table body and the latter is below it, then it's presentational
difference and should be handled outside HTML. A simple way to achieve
this would be to allow multiple <thead> elements, which can then be styled
(e.g., positioned) as desired.

The idea of using <tfoot> for summarizing information could make sense
semantically. So would the idea of giving general information about the
table data, such as explanations of notations. But keeping <tfoot> as a
vaguely defined element serves no purpose but continuity, and according to
current plans XHTML 2.0 will break continuity anyway, on purpose.
So <tfoot> should be dropped from XHTML 2.0 unless its semantics can be
agreed on, and I mean semantics that is independent of visual rendering
(though may have an impact on desirable rendering).

> Note that at least some versions of IE have a default style sheet that
> does not render thead and tfoot as running headers.  I believe this
> was a sort of quirks mode type decision, i.e. it was deliberately
> mis-implemented in order to avoid confusing authors.

No, IE just didn't implement <thead> and <tfoot> properly. I think they
have a good excuse: those elements have no well-defined semantics.

-- 
Jukka "Yucca" Korpela, http://www.cs.tut.fi/~jkorpela/