W3C home > Mailing lists > Public > www-html@w3.org > January 2003

Re: Re[2]: XHTML 2.0 considered harmful

From: Tantek Çelik <tantek@cs.stanford.edu>
Date: Wed, 15 Jan 2003 16:48:00 -0800
To: <www-html@w3.org>
Message-ID: <BA4B4116.1F372%tantek@cs.stanford.edu>

On 1/15/03 2:59 PM, "Alexander Savenkov" <w3@hotbox.ru> wrote:

> 2003-01-15T19:33:49Z Tantek wrote:
>>> If you feel that you have to use br or style attribute try to think for
>>> a second what's the reason you "need" those. Think about *why* you're
>>> trying to achieve the line break or specific styling, not *what* you're
>>> trying to achieve.
> 
>> If you cannot see the need for the style attribute, it may simply indicate
>> that you have not experienced the real world situations where it is
>> necessary.  
> 
>> I think this is in general the problem with the discussion of the 'style'
>> attribute.  On one side there are semantic purists
> Oh, I guess I'm among those. Still, "semantic purism" is not racial
> purism, and there's nothing wrong with the people who finally want to
> get ridda stylistic waste.

Right, so getting rid of presentational attributes is a good thing.

But here is the catch, the 'style' attribute is _NOT_ just a presentational
attribute.  This is a common misconception.

The style attribute is a container for CSS, whose effect is defined outside
the markup specification.  Just as <style> element is a container for CSS
whose effect is defined outside the markup specification.

The problem is that some folks who claim to or want to be semantic purists
by mistake put on semantic blinders.

> Some of them have been waiting for this
> since the early HTML drafts, some dreamed even longer.

The point is, that the style attribute is far far better than a loose
collection of "bgcolor", "valign" etc. attributes, which can sometimes be
difficult to distinguish from so-called purely semantic attributes.

The style attribute can group all the presentational annotations on an
element into one convenient, _easily_separable_ attribute that does not get
confused with all the rest of the attributes on the element.

>> that don't understand
>> what the problem is and therefore claim there is no problem, and on the
>> other side there are _experienced_ folks that have seen numerous real world
>> situations where the style attribute is not only useful, but essential.
> On the other side there are experienced folks that have seen numerous
> real world situation where other experienced and (perhaps) short-sighted
> folks decided that the cool 'style' attribute does a lot of useful
> things (no offence and nothing personal).

And there are more experienced folks who have gone through those steps,
taken the next step of moving their style attributes to style rules, and
discovered, after long experience, that there are definitely some situations
where it is very awkward (and even broken) to do so, and that using the
style attribute actually provides the best solution in those situations.

>> These real world situations have been listed in threads in this list, but
>> always ignored or belittled.
> I'm not asking you to repeat these situations here as there's been
> too much buzz around this (and we've seen some of the examples
> already). Instead an example of a system that *requires* an
> immediate upgrade to XHTML 2 would be highly appreciated.

I have no idea what that has to do with anything.  If you are saying that
there aren't any systems that require or can use XHTML 2 in its current
form, then you may be correct, and in that case, its the wrong thing to be
working on.  (see my previous comment about futile dead end science
projects).
 

>>>> Your third conclusion is one of esthetics and education. Personally I
>>>> find it much easier to think of a line than a line break, and I think
>>> 
>>> And what comes to teaching a document markup language to new people,
>>> XHTML2 looks like much simpler as a whole than XHTML1.x.
>> Then don't call it XHTML2.  Call it something else, like "SHML" (Simplified
>> Hypertext Markkup Language).
> Rather Semantic Hypertext Markup Language.

or maybe PUSHML (PUrely Semantic Hypertext Markup Language).

Tantek
Received on Wednesday, 15 January 2003 19:31:48 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 27 March 2012 18:15:54 GMT