--- Comment #4 from Vincent Hennebert <> ---
(In reply to comment #3)
> Hmm,
> In PDF 1.5 spec, TH seems exist as a workaround to the limitation of the
> html table model, from witch the tagged PDF table-model seems to be inspired.
> As I said, such table model describe a stacked rows structure, and there is
> no mechanism to describe alternative structure, where 1st column can be
> handled as header.
> Note that the "TH" scope value could be determined from the "grandfather" of
> the fo:table-cell (table-header), and/or, if implemented, the nature of the
> corresponding column or column group (header/footer column-group).

Hmmm. That makes a lot of sense. Actually the structure type of cells that
descend from a table-header is already automatically determined this way. We
could extend that mechanism to 'header' columns. A cell that is in such a
column would be given a structure type of TH with a scope or Row; A cell that
is also inside an fo:table-header would be given a scope of Both. That would
avoid manually setting role and fox:scope on every cell of the header column.

I like this idea a lot, however the time I have allocated on this is running
low. I'll see if I can modify my current implementation to do this.

So fo:table-column would take an extension property fox:header being either
true or false (the default).


> IMHO, there is no need to use a table-cell role attribute for that.
> Personally, I don't use tagged PDF, but sometimes I'm facing to the lake of
> "vertical" headers in FO table model, and I think that here, the End-user
> need is having such feature, rather than an html/TH-like tweak, breaking the
> table structure.
> And your schema illustrates perfectly what I say ;)
> *all* "row headers" are in the same column...
> And you have a cell header for each row, so with a cell implementation, you
> have to repeat the action on each row, while with a column implementation,
> you repeat it only once.

Thanks for your feedback!

You are receiving this mail because:
You are the assignee for the bug.

Reply via email to