RE: clarification of table linearization

I agree with Gregg and Jason, too.  The mention of "column order" in the
WCAT is not to be interpreted as describing _web content requirements_.
User agents may, above and beyond the minimum, provide columwise reading.
However, the fact that a table makes sense when read by columns does not
meet the WCAG test that it has to make sense when linearized by the most
elementary, bare-bones transform.  In this reference algorithm for the
content requirement, the paragraphs appear in the same order as in the
source hypertext.

[if we mention HTML it should be by way of example.  The same principle
applies to all XML and SGML applications.]

On the other hand, I would take exception to the parenthetical remark that
there are no header elements in the table to be transformed.  The algorithm
in question should be defined as a transformation on all tables, not a
limited kind of special tables.  

Linearization has limited success in keeping the sense of the content
clear.  But that success is not strictly limited to layout tables.  Some
quote data tables unquote will survive linearization just fine.  But a
larger proportion of layout tables will survive this transformation as
clear presentations than will data tables.
Is that the general understanding of the performance we expect?

Al

At 06:26 PM 1/19/00 +1100, Jason White wrote:
>I agree with Gregg. The transformation implied by the definition in the
>WCAG is one in which the table-related elements are eliminated and the
>content of each TD element (TH elements being excluded by the stipulation
>that TD cells alone can appear in a table which is used for layout
>purposes) is placed in a separate paragraph. XSL experts in the ER group
>could no doubt write a transformation that would achieve this effect,
>without difficulty.
>

Received on Wednesday, 19 January 2000 11:46:30 UTC