Re: table, cell, heading, leader, trailer, header, footer, runner, TABLE, THEAD, TFOOT, part of my review of 3.15 Tabular data

Hello all,

For an real-world example table that could benefit greatly from fixed  
thead, tfoot, tlead and ttrail elements, see all of the tables after  
the heading "Details" on this page:

<http://www.w3.org/2002/09/wbs/40318/tasks83/results>

The idea as outlined in HTML 4.01 and RFC 2854, is that, in these  
tables, the first column and row would remain fixed and the table  
bodies (as a single unit) would be presented with horizontal and  
vertical scrollers. Without these features these tables are  
inaccessible to almost everyone :-).  I'm venting a bit since I don't  
know if I've ever seen an example table that needed these features so  
bad. However, I'm also serious. These do make great real-world usse- 
case examples of why we need these features.

Take care,
Rob

Original message
-------
On Aug 10, 2007, at 2:31 AM, Robert Burns wrote:

>
> table, DOM interfaces, cell, heading, leader, trailer, header,  
> footer, runner, TABLE, CAPTION, COLGROUP, COL, THEAD, TFOOT, TR,
> part of my review of 3.15 Tabular data
>  Re:<http://dev.w3.org/cvsweb/~checkout~/html5/spec/Overview.html? 
> rev=1.78#tabular>
>
> SUMMARY
> --------------------------------
>  • consider restoring heading related and other table cell  
> attributes to HTML: @abbr, @headers, @scope, @axis,
>  • consider adding lateral runners TLEAD, TTRAIL, TC (column)  
> elements to the TABLE content model
>  • consider adding a @kind attribute to the TD element to indicate  
> whether the cell is a 'heading' cell, a 'summary' cell, or 'neither'.
>
>
> HTML 4.01 table attributes
> --------------------------------
>
> To continue to allow authors the same level of expressiveness in  
> creating tables, we need to add the HTML 4.01 table attributes to  
> HTML5. Without these attributes, HTML5 would constitute a regress  
> in terms of table semantics.
>
> @headers, has been relied upon by authors and aural UAs to provide  
> the information necessary to associate data cells and header cells.  
> However, leaving such misuse aside (which we should deprecate even  
> while retaining @headers in document conformance norms), the  
> @headers attribute is an important attribute to associate data  
> cells with header cells in more complex tables. In many cases  
> @headers can provide a shortcut for associating cells when compared  
> to @scope.
>
> The @abbr  attribute provides functionality for HTML 4.01  
> conforming documents that cannot be achieved with an HTML5  
> conforming document.  There may be an easier or better way to  
> abbreviate headings (including having UAs rely on UA abbreviation  
> dictionaries), however authors may still need to provide  
> abbreviated headings when no such dictionary entry exists.
>
> While the @scope attribute is included on the TH element in the  
> current HTML5 draft, it has been removed from the TD element. The  
> use of @scope on the TD element is important to be able to make a  
> data cell behave as a heading for other data cells. The existence  
> of a @scope attribute on a TD element indicates that the data cell  
> contains a heading for other data cells. It may even be useful to  
> allow an explicit value for  @scope of 'auto' or 'all' to indicate  
> a data cell contains a heading, but should not have its scope reduced.
>
> The @axis attribute allows authors to designate dimensions beyond  
> the natural two dimensions available on a table. For some  
> relatively complex data, the use of an axis can vastly simplify the  
> visual presentation of  the data when compared to other approaches.
>
> All of these attributes address well known use-cases as outlined in  
> the HTML 4.01 recommendation [1]. It is important to start from use- 
> cases and problem statements before proposing sweeping changes to  
> the web like this. In order to remove these attributes it is  
> therefore important for us to discuss the problem-statement that is  
> being addressed by removing these attributes. In other words, what  
> problem will removing something like @headers solve?
>
> Lateral runners: TLEAD and TTRAIL
> -----------------------------------------------
>
> THEAD and TFOOT are intended as column global headings and column  
> global summaries: appearing fixed and available for visual review  
> while scrolling the table bodies. The rows currently offer no  
> similar mechanism. To allow tables to work in a more symmetrical  
> way, consider adding table leader (TLEAD ) and table trailer  
> (TTRAIL) elements. These would be lateral runners and work in a  
> similar way to the already existing top and bottom runners, THEAD  
> and TFOOT. However, these elements would comprise columns rather  
> than rows., so we should consider adding a content model too TLEAD  
> and TTRAIL of one or more TC table column elements. The TC element  
> content model would be the same as a TR. However, the arrangement  
> of cells within the TC would be vertically stacked cells rather  
> than horizontally arranged cells. The TTRAIL and TLEAD elements  
> would constitute their own column groups: and would have the same  
> number of rows (TD and TH cells) as the table body or table bodies  
> taken together (or implied if the number of rows in the table  
> exceeded the number of cell in the TLEAD or TTRAIL).
>
> Table summary cell
> --------------------------------
>
> To provide a way of marking a cell as a table summary cell,  
> consider adding an boolean attribute @summary. It may even be  
> useful to add a @kind or @type attribute to TD cells to be able to  
> mark them as either headings or summaries (i.e., kind=''heading' or  
> kind='summary'). Setting a cell to be a summary cell would have a  
> corresponding algorithm like the headers association algorithm but  
> working in the opposite direction (up and to the left rather than  
> down and to the right).
>
> Summing up
> --------------------------------
>
> these proposals are to round out the expressiveness of tHTML. The  
> newly proposed elements would be incompatible with existing UAs in  
> the text/html serialization. They might be more likely to degrade  
> gracefully within the XML serialization, ,however non-HTML5 UAs  
> would not know how to present them. CSS also does not support the  
> lateral runner concepts proposed in this review, so its table model  
> would have to be updated to support those runners. However, CSS  
> also needs enhancement to handle the existing THEAD and TFOOT  
> runners, as authors may want to provide fixed THEAD and TFOOT with  
> the main body of the table presented in a fixed frame with overflow  
> and clip control through scrollers. I do not believe this level of  
> control is available now with CSS or proposed with CSS3 even for  
> the existing table runners. elements with scrolling table bodies.  
> So adding lateral runner support to CSS could occur at the same  
> time CSS is enhanced to support top and bottom runners.
>
> [1]: HTML 4.01 Chapter 11: Tables <http://www.w3.org/TR/html4/ 
> struct/tables.html#h-11.3.2.1>	
>

Received on Wednesday, 15 August 2007 08:25:57 UTC