Re: Parsing problem with misnested tags

Philip TAYLOR (Ret'd) wrote:
> You may be correct (I'm referring to the "significant
> number of sites" here), but is there any evidence to suggest
> that an approximately equal number of sites do not assume
> exactly the converse ?

Yes: the fact that every single browser the site author might have 
tested with gives the same rendering.  If the site author wanted a 
different rendering, he would have changed the markup to produce the 
desired rendering.

Now there are some sites where the rendering is just broken in all UAs, 
and where we have to assume the author didn't bother to look at their 
own site (or couldn't for whatever reason).  But usually authors would 
tend to look at their page once done authoring.

> It is certainly not clear /a priori/
> that the italicisation should continue, nor that that was
> the intended behaviour.

I agree that that would perhaps be the case if UA behavior differed on 
this point (though even then, I'd go with the majority UA rendering as 
what sites intend).  But it doesn't.

The other evidence, of course, would be from bug databases.  Those are 
open for Gecko and Webkit if someone wants to do a bit of searching.

>  From a purely personal perspective, I believe that
> handling real-life web content that is /wrong/ is
> at the very tail of the distribution of the criteria
> that we should be considering.

You're welcome to this personal perspective, as long as you realize that 
most web content is "/wrong/".

-Boris

Received on Wednesday, 12 November 2008 21:49:26 UTC