Re: ISSUE-158: object-content-model - Chairs Solicit Alternate Proposals or Counter-Proposals

Edward O'Connor, Wed, 14 Mar 2012 15:15:04 -0700:
> I've written an alternate Change Proposal for ISSUE-158:
> 
>     Preserve the transparent content model of the object element
>     http://www.w3.org/html/wg/wiki/User:Eoconnor/ISSUE-158


The CP’s “negative effect” is a straw man:

  "Web authors will not be able to use <object> to work
   around the content models of other elements."

Someone who wants to go against <object>'s content model by making its 
fallback render as the main content, does not need my CP for that. May 
be my CP makes it a little bit more tempting to try. But that's all.

Only if an author comes up with fake values - for the required @data or 
@type - that -firstly-  passes the validator and -secondly- makes the 
fallback 
render as main content, can we talk about someone who uses the effect 
of my CP to 'work around' content models.

You mention the spec's foreword. And the example there - a button with 
a textarea inside - does indeed seem 'highly confusing'. I understand 
that you imagine that one could wrap an <object> around that <textarea> 
to 'work around the content model'.  But for the author to do that, 
would require him or her to - either - make the fallback render by 
default - or - to misrepresent whatever <object> was embedding: A 
textarea as fallback for embedded content - when would not make sense 
unless the <object> contained e.g. a flash based form. And in that 
case, it would probably not be kept inside a <button> in the first 
place.

Many 'highly confusing' possibilities have not been forbidden by HTML5 
in a syntactical way. For instance, precisely <object> could be used to 
contain e.g. a flash based application or an interactive svg: Such a 
thing would be valid, even if it was contained inside a button - or an 
anchor and despite its potential for confusion. I would not mind if it 
was possible to say that the form example was syntactically invalid. 
But regardless: The bizarre example in the foreword, should not be held 
against a sensible example of a link, e.g. to a longer description, of 
the embedded object. 

Because my CP is not about making it possible use <object> to 'work 
around' anything, it is also not going to be confusing: The world's 
many HTML 
tutorials have focused very little on whether the parent of <object> is 
a <p>. On the contrary, there are numerous examples of <object> with a 
<p> in the fallback. Including examples where the parent is a <p>. The 
reason it is so easily forgotten, is probably because one perceives the 
fallback as another, sealed off, layer.
-- 
Leif Halvard Silli

Received on Thursday, 15 March 2012 06:20:13 UTC