Re: Adjusting the definition of fallback-free object

On Tue, 07 Apr 2009 16:27:01 +0200, Simon Pieters <simonp@opera.com> wrote:

> This page
>
>    http://www.systemrequirementslab.com/bonus/t/analysis2.tmpl?registration_option_id=7840
>
> fails to work given these rules
>
>    http://www.whatwg.org/specs/web-apps/current-work/multipage/dom.html#dom-document-nameditem
>
> (...which WebKit and Opera more or less implement. Firefox and IE do  
> things differently.)
>
>
> I think it could be made to work without regressing other pages by  
> changing the definition of "fallback-free" such that an object is  
> fallback-free if it does not have any object or embed descendants. Maybe  
> applet should be in the list, too. img should probably not be in the  
> list considering spacer gifs.

I've looked at http://philip.html5.org/data/object-with-id-or-name.xml.bz2 (don't open that in a browser) and noted the following:

  * Most have just an embed nested in object.
  * Some have nested objects. Many of those have one of the following nested in the inner object:
     - "FAIL (the browser should render some flash content, not this)." or other copy-paste cruft
     - "<a href=...><img alt="Get Adobe Flash player" src=...></a>"
     - Actual alternative content (e.g. an image map)
  * None of them had an <applet> nested in <object>.
  * None of them had an <iframe> nested in <object>.
  * A few have <form> (with id/name) nested in <object>.

In conclusion, I think the list should be just object and embed.


> Discussion at http://krijnhoetmer.nl/irc-logs/whatwg/20090407#l-489
>
> # [16:03] <zcorpan> the page does <object id=ie><comment><object  
> id=moz><table><tr><td>&nbsp;</table>
> # [16:03] <zcorpan> along with if (null != document.moz) ... else if  
> (null != document.ie) ... else return // error
> # [16:03] <zcorpan> both objects are ignored because they have "fallback"
>



-- 
Simon Pieters
Opera Software

Received on Wednesday, 8 April 2009 07:11:43 UTC