W3C home > Mailing lists > Public > w3c-wai-hc@w3.org > October to December 1997

survey of TABLE issues

From: Al Gilman <asgilman@access.digex.net>
Date: Fri, 10 Oct 1997 00:52:38 -0400 (EDT)
Message-Id: <199710100452.AAA23481@access2.digex.net>
To: w3c-wai-hc@w3.org (HC team)

Raman's edit of my "operational concept" is still a pretty good
baseline, or rough statement of where we are, for what it decides.

I think that the capability to tie a table to some sort of a
document that gives more information about the data discipline
that is behind the table is something that has a high likelihood
of being part of the answer, if we can work out the
implementation.  A minority will use it, but for those who are
interested in providing web pages which are competent to transfer
database fragments, it is a necessary option.  People writing
tables out of database systems will not find it hard to provide
some dictionary support.  We should at least have the capability
to refer to it, and tie a particular "dictionary" specifically to
an individual table.  It is too likely that several tables in one
document will be drawn from different databases each with a
different data model.

I have little doubt that the IG would fail to back up the
desirability of such a capability, so I want to get to work with
the browser people and the RDF people to work on implementation
options.  [more on this in a separate mail].

Dave and I had a good talk today.  This is an attempt to reflect
points of agreement and non-agreement from that talk.

For a browser to be able to present the contents of a table
as a hierarchical list is a valuable function.

	- sometimes (to most of the time) you can do this
	with existing tables without the need of additional 
	markup such as the AXES, AXIS, ABBR in the current

	- adding that markup will increase the effectiveness
	We don't really know how much.

The "probe a single cell" scenario matters.  It is desirable
for the table to make clear what other cells are closely
related to the current cell and how, enough so that a CSS
macro could compose a sentence that give the complete message
of the cell [content value] including what it needs to carry
along from the context provided by the table.

The biggest area of confusion has to do with how one finds the
other cells related to one "anchor" cell needed to provide a 
complete statement of what the "anchor" cell tells you.  

Dave feels the AXIS/AXES markup does this.  I am still not

We agreed to seek examples from life for analysis to aid the

There is a representative of the list-of-records table at


for which there is a straight text alternate at


in case you can't follow the table version.

This is the highly guessable case where the first column is the
key or "what else in the row you always want to read with any

I have not yet located a page where it takes multiple columns
to form a sufficient key.

This table is interesting in that it uses TD markup throughout
for cells that are field descriptors and cells that are field
values without distinction.

I had almost started to believe that I could interpret an AXES
link from a TD to a TH as an IS-A relationship and an AXES link
from a TD to another TD as a KEY relationship.

But this example reminds me that just as some people will put the
first column in as all TH's for cosmetic reasons even 'though
that column contains data below the top row or so, there are
others who will produce tables where what I would call a field
descriptor is in a TD element and not a TH element.

I don't think we can trust that TD and TH fields will be used for
data values and field descriptors respectively, with any real

For the record, there is another table I want in the evidence.
This is found at


This illustrates a table that I would give a class of SET.  If
you state that the semantic structure of this collection is at
the SET level, then the fact that it has been spread out in a
table for cosmetic purposes will not mislead one into thinking
that there is any deep order or connectedness semantics between
neighbors, rows, or columns.

...Too late.

More examples, more organization tomorrow.

-- Al
Received on Friday, 10 October 1997 00:52:55 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 19:56:11 UTC