Re: ISSUE-155 counter proposal

Tab Atkins Jr., Wed, 23 Mar 2011 09:00:36 -0700:
> On Wed, Mar 23, 2011 at 7:03 AM, Leif Halvard Silli:
>> Tab Atkins Jr., Wed, 23 Mar 2011 06:25:13 -0700:
>>> On Wed, Mar 23, 2011 at 3:13 AM, Leif Halvard Silli wrote:
>>>> Ian Hickson, Wed, 23 Mar 2011 04:06:47 +0000 (UTC):


>>> The point is that non-CSS UAs can (and, apparently, *should*)
>> Were is the "apparently" coming from?
> It seems apparent that non-CSS UAs should display a border on tables
> automatically.

Again, according to Ian [1], Webkit, Opera, IE, Firefox are non-CSS 
user agents when CSS is disabled. And they do not live up to your 
border ideal. And neither do text browsers (e.g. w3m). So it evidently 
does not seem apparent to any UA vendors, at least.

> This latter one.  It's well-established that CSS UAs shouldn't display
> a border by default (changing that now would likely mess up sites that
> don't expect the table to suddenly get slightly wider).  But non-CSS
> UAs should default to always displaying a border, unless they have
> some alternate way to apply styling besides CSS.

That you value the usefulness of borders in a table, is appreciated. 
But layout tables is a technique that is older than CSS. 

>> ALSO: Can you point to a place in HTML5 which claims the same thing
>> that you claim?
> What am I claiming that needs justification?

The claim that tables *without* the border attribute should be rendered 
with a border in non-CSS user agents. In an ideal world, I admit that 
to have borders by default fits well with the idea that tables should 
be data tables. And I would much have preferred it that way. But don't 
quite dare to dream about such a big change ...

>> NOTE: I have a vague feeling that you and Ian are of the opinion that
>> if a user agent support the border attribute, then it support CSS, only
>> not with the correct syntax. Am I on to something?
> No, that's not what I'm assuming.  (Though, visual HTML UAs pretty
> much all support CSS, so it may be considered trivially true.)
> My goal is to not require authors to use presentational markup solely
> for low-functionality UAs, when those UAs can instead change their
> default rendering and make tables work for *all* users.

Sounds like one must detail-govern how 'low-functionality user agent 
vendors' define their products. A low-func UA which is able to hide and 
show table borders by looking at the border attribute, why would it 
want to remove that functionality?

And I guess that disabling CSS in a high-func browser would also render 
the borders?

>  I want to
> avoid a situation where only the subset of authors who know to add
> border="1" to their tables get decent rendering.

The goal is good.

> Of course, right now no one gets decent table rendering in many
> non-CSS UAs.  This is an indication that this is probably not a very
> important issue in the first place.  If it is sufficiently important
> to care about, though, it should be addressed in a way that
> automatically solves the issue for everyone.

leif halvard silli

Received on Wednesday, 23 March 2011 17:27:37 UTC