RE: [techs] Definitions of complex data tables

I certainly agree with Sailesh that tables with multiple header rows and/or
header columns (and with row and/or col spans) are complex and difficult to
comprehend with a screen reader. However that is not the definition of
complexity that interests me. I want the question to be, for what class of
tables are TH (and scope) inadequate to present the headings structure of
the table so that the headers/id markup would be required. Those are the
tables I want to be called complex. We can leave the word "complex" out of
the requirement and paraphrase from Sailesh's message with:

 

Headers/id markup is required for tables in which heading cells are
associated with data cells that are not exactly below them or exactly to
their right. (I don't like the wording either!)

 

As Sailesh knows from the tool he works with, when all heading cells for a
data cell are found in data cell's row and column (an l-r system) then the
headers/id markup, if you want to add it, can be algorithmically generated
knowing the heading rows and columns, in effect, knowing the TH's. That is
exactly why I think headers/id markup should not be required in these cases.
Since the headings information can be algorithmically generated, I think we
should leave it to screen readers (assistive technology) to figure implement
the algorithm.

 

I still like the idea of requiring summary or CAPTION for data tables and
summary="" for layout tables. 

 

Jim

Accessibility, What Not to do: http://jimthatcher.com/whatnot.htm.

Web Accessibility Tutorial: http://jimthatcher.com/webcourse1.htm.