My organization has an internal table model, based on an abstract
multidimensional data array. It's great in theory for representing things
(like experimental data with 27 independent variables). But we also support
the CALS tabel model, not because it's the most worderful thing in an
intellectual sense, but because (1) there are tools for editing CALS tables
and (2) it's comprehensible to editors and compositors used to dealing with
tables as a two-dimensional way of presenting data.
We're also willing to accept similar compromises with mathematics; our staff
are used to markup languages (e.g., dialects of eqn and TeX) that compromise
between logical purity and reasonable formatting.
But, returning to Terry's other point, we need to be very careful about how
much we allow formatting to impinge on our markup application. I know I bring
this up at every available opportunity, but I'll say it again: it's much
better to capture some information relationships than to try to build a new
formatter. What's making HTML complex is attempting to add thing like video
attributes, not building a basic system of sections, subsections, paragraphs,
lists and--yes, tables and math. If you want pretty, author-controlled
presentation, follow Netscape's lead and transmit PDF.
Dr. James D. Mason
(ISO/IEC JTC1/SC18/WG8 Convenor)
Oak Ridge National Laboratory
Information Management Services
Bldg. 2506, M.S. 6302, P.O. Box 2008
Oak Ridge, TN 37831-6302 U.S.A.
Telephone: +1 615 574-6973
Facsimile: + 1 615 574-6983
Network: masonjd @ ornl.gov