Re: Presentational markup

Anne van Kesteren, Thu, 25 Mar 2010 11:18:56 +0100:
> On Thu, 25 Mar 2010 00:21:50 +0100, Sam Ruby <rubys@intertwingly.net> wrote:
>> On 03/24/2010 06:49 PM, Anne van Kesteren wrote:
>>> On Wed, 24 Mar 2010 23:32:08 +0100, Sam Ruby <rubys@intertwingly.net>
>>> wrote:
>>>> One of the things that attracted me to HTML5 was that it purported to
>>>> document the web as it exists as opposed to other efforts at the W3C
>>>> which intended to change the web to their vision as to how it should be.
>>> 
>>> Can't we have a mix of pragmatism and idealism?
>> 
>> Could we?  Yes.  But my observation is that what we tend to have 
>> here is a situation of what I call "conjugating adjectives".  You 
>> know how in English one says "I am", "you are", "he is", ...?  What 
>> I observe is people who have a tendency to take this a step further 
>> and go "I am idealistic", "You are dogmatic", "they are orthodox".
> 
> What I meant is that HTML5 is not completely about the Web as it 
> exists. It still sets goals. Instead of describing the various ways 
> in which implementations are different from each other, it sets a 
> path for how they can converge over time, adjusted along the way as 
> appropriate. I think this is important in order to make progress.

That user agents converge about a documented way to interpret HTML, 
will save resources: Vendors can compete about the best implementation, 
rather than competing about supporting different features. This will 
also save resources for authors, who will find it simpler to write code 
that works cross browser. Even if they don't validate, they should 
experience this progress.

Problem: I have heard 'saving vendor resources' as argument for 
removing certain features from HTML. (Example: @coords in the A 
element.)

> Likewise setting goals for content authors make sense to me. 

The suggested rules for use of xmlns is also about setting goals for 
authors. Why doesn't that make sense to you then? 

What are content authors' role in the progress of the language?  From 
what you wrote above, it seems like progress of the language is only 
about getting user agents to agree. I think authors are also crucial 
for the future of the language.  So what does it mean to set goals for 
content authors? And what relationship should there be between what 
user agents converge about and what authors may do?

Example: the <meta> content-language element. My change proposal allows 
authors to use <meta> c-l to make sure that @lang works as HTML5 
defines it in existing user agents. The current spec does not. 

Example: Advice to e.g. wrap <section> inside <div> or opposite, to be 
compatible with existing UAS, also really is about long-run goals for 
the language. But why is it a *goal* that I as author stop using 
<object> without placing a @data or @type attribute inside? Does it 
help anyone if I begin to place an empty @data there? I think authors 
are willing to wrap a div around <section> because the goal of doing so 
is easy to understand. Some other things are harder to understand, if 
at all possible.

User agents converge. Authors are restricted from taking advantage of 
all of this progress. Despite that some of the features that user 
agents have converged around (e.g. - currently - the removal of 
a@coords ) might have been justified by an evaluation of how much (read 
little) authors (for various reasons) made use of certain features ... 

> Of course we need to evaluate over time whether they can be met, 

The first thing to evaluate is whether the goals are logical and good. 
We should do that before counting support.

> but HTML5 just started to get some traction.

You mean the current authoring requirements?

> Transitional has been the 
> line of designer/developer advocacy for over a decade; I do not think 
> we should be surprised there is plenty of it on the Web. HTML5 offers 
> a break from this

A break from "this"? No. In HTML4/XHTML1, there is an unfortunate and 
illogical link between quirks mode and certain features. HTML5 betters 
this in some respects - there is no doubt about that. But with regard 
to other features - legacy or what not - then HTML5 seems to be 
indifferent about whether I use quirks mode or standards mode. If (to 
take that same example again) I want to use <strike>, then I can just 
as well skip using the entire doctype - it is all the same to HTML5.  
Result: Better to use transitional doctype, as before.

> and given the reactions of designers/developers so 
> far, it seems to be positively received. I've yet to see blog posts 
> asking for the return of the <font> element or cellspacing="" on 
> <table>.

Google doesn't complain/care ... 
<!DOCTYPE html> looks cool, regardless ...
Authors continues to use Transitional.
-- 
leif halvard silli

Received on Thursday, 25 March 2010 13:21:06 UTC