Re: function and impacts (was: @scope and @headers reform)

James Graham wrote:
> That suggests that the automatic association should only occur if the 
> header has
> the a row/colspan (depending on the direction we are looking along) that 
> is
> greater than or equal to that of the cells it is being associated with.

James Graham wrote:
> Apparently this was a case that Simon and Ben discussed for the original
> smart headers algorithm that got lost along the way somewhere. I've 
> changed
> it in the table inspector now so that it is handled correctly.

For any interested archeologists, there was discussion about this on 
#whatwg:

<http://krijnhoetmer.nl/irc-logs/whatwg/20080925#l-601>

Some table accessibility research and discussion took place spontaneously 
earlier that day:

* <http://krijnhoetmer.nl/irc-logs/whatwg/20080925#l-322>

I do recall Simon 'zcorpan' Pieters and I looking at various patterns of 
spanned cells that we found in real tables:

<http://krijnhoetmer.nl/irc-logs/whatwg/20080925#l-710>

Not sure if this particular observation ever went past an MSN coversation, 
so that might be why the prototype didn't have it. Good work, James. :-)

Gez Lemon wrote:
> That's very impressive. I've just tested it with an example that
> doesn't use the headers attribute at all [1], and the associations are
> all correct. I'd like to test it with a wider range of tables, but the
> algorithm satisfies the examples I put forward. If nested headers are
> allowed in HTML5, I think that algorithm could work very well.

Yes, that's the idea of it. :-P

I'd like the HTML5 algorithm to align more closely with Smart Headers over 
the coming years. Also, for the Smart Headers algorithm to continue 
receiving improvements based on observing the patterns in real tables and 
testing how well they are supported by it.

-- 
Ben 'Cerbera' Millard
<http://projectcerbera.com/web/study/>

Received on Friday, 26 September 2008 05:55:05 UTC