[whatwg] Quirks mode handling of rowspan="0"

[Ian Hickson]

On Sun, 2 Sep 2012, Boris Zbarsky wrote:
>
> Per HTML5 spec, rowspan="0" should span to the end of the table section.
> 
> WebKit seems to not implement this at all.
> 
> Opera implements this in quirks and standards mode.
> 
> Gecko implements this in standards mode only; in quirks mode, 
> rowspan="0" is treated like rowspan="1".  When we last tried to do the 
> "right" thing in quirks mode, back in 2003 or so, we ran into site 
> compatibility issues, hence the current setup.
> 
> I don't know what Trident does here; worth testing.
> 
> In any case, my suggestion is that in quirks mode, rowspan="0" not be 
> supported.

On Mon, 3 Sep 2012, Kang-Hao (Kenny) Lu wrote:
> 
> IE seem to follow WebKit in every mode I tested with.

(i.e. not support)


On Thu, 6 Sep 2012, Simon Pieters wrote:
> 
> greping for "rowspan=\"0\"" in
> http://www.paciellogroup.com/blog/2012/04/html5-accessibility-chops-data-for-the-masses/
> and http://dotnetdotcom.org/ I find the following pages being broken in Opera
> but working in Firefox/Chrome:
> 
> http://www.persianv.com/
> http://www.quicherchetrouve.be/index.php?r1=2&an=3&anb=239&page=3
> http://www.kvaak.fi/keskustelu/index.php?topic=1914.0
> http://www.timekillerarcade.com/game.php?id=1941
> http://www.elginisd.net/cgi-bin/calcium37.pl?CalendarName=David_Schmitt&Op=ShowIt&Amount=Week&NavType=Both&Type=List&Date=2008/8/1
> 
> I stopped after going through about half of the matches. All of these 
> are quirks mode. I didn't find any that work in Opera/Firefox but are 
> broken in Chrome (i.e. rely on it being unsupported in standards mode). 
> I didn't find any that work in Opera but are broken in Firefox/Chrome 
> (i.e. expect it to work in quirks mode).
> 
> This is enough to convince me that the feature should not be supported 
> in quirks mode.

Done.


On Mon, 3 Sep 2012, Jukka K. Korpela wrote:
> 
> Generally, attempts at defining "quirks mode" would mean making it an 
> alternate mode and will not be successful due to the wide variation 
> across browsers and versions. It's called "quirks" for a reason.

We have indeed created an alternate mode (two, actually). See the HTML 
spec, search for "quirks".


> Specifically, as some browsers already support rowspan="0" in quirks 
> mode, and some don't, you cannot ensure backwards compatibility no 
> matter how you define it.

Depends if UAs are willing to change quirks mode. By and large, they are.


On Mon, 3 Sep 2012, Boris Zbarsky wrote:
> 
> Yes, it does: once all browsers agree, the barrier to entry for new 
> entrants is lower because they just need to implement the behavior 
> everyone already agrees on.

That's the dream!

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