> I decided against separate header and footer syntax because I wanted to
> keep the cell/row data format homogeneous and independent of the table
I was interested in a way to create tables with vertical headers, or
even with both horizontal and vertical as would be typical for such
things as cross tabs, pivot tables, contingency tables, and matrices.
These are all common and popular table layouts so it would be useful
to be able to represent them.
Rather than the first idea of using a character to flag a header cell,
here's an alternative that might be more in line with your design
goals. How about adding a "headers" option that could be included in
the bracketed definition. Ex:
.Sales by Region
| |West |Central |East | Total
|Q1 |270 |292 |342 | 904
|Q2 |322 |276 |383 | 981
|Q3 |298 |252 |274 | 824
|Q4 |344 |247 |402 | 993
|Total |1234 |1067 |1401 |
In the example above (those are ones, not ells), both the first row
(1r) and the first column (1c) would be headers. This syntax would be
backward compatible as well as allow for a good deal of expansion.
For instance if you wanted the second column, you could use "2c".
Negative indexes could be used to count from the other side of the
table, so "-1r" would be equivalent to a footer, and -1c would be the
rightmost column. So perhaps [headers="1r,1c,-1r,-1c"] might be
desired for the table above.
This could further be expanded with cell level syntax, for example the
empty cell in the upper left corner could be specified with "1.1", and
the blank cell in the bottom right would be "-1.-1". Finally negation
could be added with a ! to invert the current setting. So above if
you wanted header and footer rows, as well as the left and right total
columns, but not the two empty cells, you could use [headers="1r,
Ok, maybe I'm getting carried away. Perhaps just file this away for