Re: Initial approach to the reading order issue

I am aware of no reason why basic tree transformation controls, of the
kind which I have described, could not be added to CSS 2. There are
already visual flow objects. Tree transformation takes the concept one
step further by allowing the structure of the document to be modified
for the purposes of generating optimal output in a particular medium. I
have also argued that, given the intention of the CSS working group to add
support for footnotes and end notes, such transformations will be
required, or at least desired in any case, particularly for the audio and
braille rendering of notes, but also to some extent as a means of
regulating their placement in a printed text.

Al has raised several interesting points. Firstly, does there need to be
an attribute which can indicate whether a table is best understood by
reading down the columns or across the rows? This would at least allow
columnar material to be identified as such. However, I would not support
any strategy which condones the use of the HTML TABLE element for any
purpose other than the representation of tables. Although some authors
misuse HTML tables, the WAI should not endorse such a practice by
attempting to compensate for it in the markup. Control over the layout of
documents is the province of style sheets, and will become even more so as
CSS is more widely adopted.

Received on Tuesday, 30 September 1997 20:33:49 UTC