[whatwg] Publishing another Web Forms 2 Call For Comments soon

[Ian Hickson]

On Sun, 16 Jan 2005, [ISO-8859-1] Olav Junker Kjær wrote:
> 
> The real problem is that there is no way to declare that all form 
> controls in a single table row is part of the same dataset and should be 
> submitted (and updated) as a unit, since HTML does not allow TR elements 
> to be contained directly by a FORM. This is a serious problem, since 
> tabular data which you want to edit "row by row" is exactly the type of 
> data TABLE was intended for.

True. Authors will effectively be forced to have a <form id="rowNNN/> in 
the first column of each row, and then have form="rowNNN" on each form 
control. Not perfect, but we can't really change the <table> content model 
(table parsing is already a house of cards, making it worse is a legacy 
content minefield).


> But maybe its just because I understand the semantics of the form 
> element incorrectly? I think of a form as as a grouping of related input 
> fields which is edited by the user as a whole. By this understanding, it 
> makes sense to have the form defined by containment. OTOH, if the form 
> element is thought of more like the data model in XForms, then it makes 
> sense to define the FORM separately from the input widgets, and allow 
> widgets to bind to several data models. I dont think this fits too well 
> with how forms usually are thought of in HTML, though.

The beauty (?) of the Web Forms 2 definition is that you can use either 
model, as you see fit.

-- 
Ian Hickson               U+1047E                )\._.,--....,'``.    fL
http://ln.hixie.ch/       U+263A                /,   _.. \   _\  ;`._ ,.
Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'