[tei-council] @rows="0", @cols="0"?

[Martin Holmes]

I'm not sure I understand Gaby's example below, but I have a feeling 
that if the intention is to say that it's cell B that's missing from row 
2, and cell C that's missing from row 3, then I think empty cells would 
be included; as far as I can see, if the second cell spans both columns, 
it somehow constitutes both of them.

Next question: given that we mostly seem to feel that values of zero are 
wrong, would we go so far as to forbid them through Schematron (given 
that every Schematron check adds to build time and code complexity), or 
should we just quietly allow people to perpetrate zeroes without sanction?

Cheers,
Martin

On 13-02-02 10:04 AM, Gabriel Bodard wrote:
> I feel as though the answer to this ought to be yes (even if only in a
> hypothetical, never-really-going-to-happen kind of way) but I'm not sure
> I can think of even that. Trying to twist my mind...
>
> Would you ever get a case when a table (in the proper sense of a series
> of rows, each cell in which lines up semantically with a column heading
> above) has both spanning cells and absent cells (that would need to be
> so marked)?
>
> So if the table has columns A, B, C, and rows 1, 2, 3, with a structure
> something like:
>
> <table cols="4">
>     <row>
>       <cell role="label">A</cell>
>       <cell role="label">B</cell>
>       <cell role="label">C</cell>
>     </row>
>     <row>
>       <cell role="label">1</cell>
>       <cell>1A</cell>
>       <cell>1B</cell>
>       <cell>1C</cell>
>     </row>
>     <row>
>       <cell role="label">2</cell>
>       <cell cols="2">2A</cell>
>       <cell>3C</cell>
>     </row>
>     <row>
>       <cell role="label">3</cell>
>       <cell cols="2">3A</cell>
>       <cell>3B</cell>
>     </row>
> </table>
>
> How can you tell from the markup (given the content is arbitrary) that
> row 3 does not have the same structure as row 2? Would a <cell
> cols="0"/> in row 2 make it clearer that there is no value for column
> "B" in that row?
>
> That's the best I can do. I'm not sure even I'm convinced.
>
> G
>
> On 01/02/2013 18:33, Martin Holmes wrote:
>> I'm working on this ticket:
>>
>> http://purl.org/TEI/FR/3511398
>>
>> where the objective is to express any constraint on an attribute value
>> which is currently only expressed in a valDesc as Schematron so it can
>> be enforced. So far the only candidates I've found for this are @rows
>> and @cols in att.tableDecoration. These are data.count, which permits
>> zero, but the valDescs would seem to imply that zero is wrong.
>>
>> Can anyone think of a scenario in which it would be reasonable to have:
>>
>> <row cols="0">
>>
>> or
>>
>> <cell rows="0">?
>>
>> Cheers,
>> Martin
>>
>