NW posted some corner cases I sucessfully tested my code
against (see below)
Does your code handle it okay? like to post ? :)
Sadly mine doesn't handle the test case - the cell 'E' basically trips
it up because of the need to generate a cell in column two to ensure 'E'
is in column three... (to be honest in my opinion the table doesn't
*deserve* to rendered correctly because it's duff markup :)
I did post a table normalization stylesheet that may be of interest to
you a while back that's available from Dave P's faq
(http://www.dpawson.co.uk/xsl/sect2/N7450.html#d8897e730) which converts
rowspans and colspans to individual spans. I would think you could
modify that to generate the 'missing' cells in the test case and then
apply a non-recursive table formatting stylesheet to it as part of a
two-stage transform - it would certainly be faster than 10 minutes ;)
Cheers
andrew