W3C home > Mailing lists > Public > www-html@w3.org > April 2007

Re: HTML5 script start tag should select appropriate content model according to src

From: Henri Sivonen <hsivonen@iki.fi>
Date: Wed, 25 Apr 2007 11:49:54 +0300
Message-Id: <E748608B-91A2-49B0-81B4-9017F9BDBDB4@iki.fi>
Cc: "Patrick H. Lauke" <redux@splintered.co.uk>, www-html@w3.org
To: tina@greytower.net

On Apr 24, 2007, at 23:37, Tina Holmboe wrote:

> On 23 Apr, Henri Sivonen wrote:
>
>>>   No, it is part of the presentation. The acid test apply: if you
>>>   remove the italics, will the content /still/ be "Latin"? If
>>>   yes, then the italics is presentational, and vice versa: if you
>>>   make a word italics, is it then also Latin? If not, then it
>>>   isn't structural.
>>
>> If italics didn't carry any signal, why would authors use italics?
>
>   Italics is a way to *represent* semantics - it is not semantic
>   in itself - unless you define it to be.

The point is that it is practical to encode in a file that a run of  
text is requested to be rendered in italics. It is less practical to  
encode the profound reason why.

> That's why giving an
>   element which was previous defined as "no semantics, lots of
>   presentation", a semantic interpretation.

I think that sentence might be missing something.

However, at the end of the day, what reasoning consumers can do based  
on markup doesn't depend on what the spec says but on how markup is  
actually used by markup producers.

>   I /still/ want to be able to write data mining tools which are  
> able to
>   tell the difference between <latin name for cat> and <italic font>,
>   and the WA1 is not helping.

If you have a real implementation need in mind instead of mere  
illustrative situation, it makes your case a lot more interesting.  
Why do you want to be able to write data mining tools that recognize  
taxonomical names? Who would use the tools and for what purpose?  
Given that taxonomical names could be enumerated in a dictionary  
against which content could be compared, why do you believe that  
explicit markup would better serve your use case than a dictionary- 
based heuristic?

>    * If there is no use case for <latin name of cat> because people  
> have
>      no interest in marking documents up that way, then there is
>      similarly no use case for <i> to be used for the same purpose,  
> and
>      we can safely toss it out as presentational.

You assume that it is OK to toss out presentational stuff categorically.

>    * If, however, there IS a use case for marking up <latin name of  
> cat>
>      then there is no point in using <i> for the purpose, since it has
>      in the past /not/ been used for it, but rather for presentation.
>      The amount of truly amazing cat names that would result is
>      staggering.

Agreed. However, it would make sense to use <i> with a microformat  
class.

>    * If <i> /had/ been used, hopefully consistently, to mean <latin  
> name
>      of cat> in the wild, and in the past, then there would be use  
> case
>      for the tool I described (yes! I, for one, want to be able to
>      automatically extract information from documents. I'm not alone.)
>      *but it will require unambiguous and semantic markup to function.

That wish is too generic to constitute a serious use case.

>    * <i> /has not/ in the past been used to mark up <latin name of  
> cat>
>      in anything but very rare, exceptionally rare, cases. A tool  
> cannot
>      tell "something which is italic" and "something which is the  
> latin
>      name of a cat" apart - even if WHATWG says they should.

A tool with a dictionary of taxonomical names could.

>>>   that specify it to have a semantic meaning which people, again
>>>   according to yourself, have no interest in using?
>>
>> To sprinkle disguising semantic pixie dust to sooth the concerns of
>> anti-presentationalists, I guess.
>
>   That's a peculiar comment. How does it relate to the  
> specification of
>   a structural language, and what's a "anti-presentationalist"?

It relates to a structural language in the way that the definition  
allows a useful presentational element to masquerade as something  
that can be included in a language that purports to be purely  
structural. Personally, I'd prefer to just say that some  
presentational elements are included.

An anti-presentationalist is a person who objects to presentational  
features.

>   It's not good
>   enough. It has the entirely wrong focus, has started off on the  
> wrong
>   foot, is huge, unwieldy, and - my personal view - badly written.

I'm looking forward to your reply to Lachy about these points.

>   At this point in time I suggest we start with 4.01 Strict, toss out
>   deprecated elements and *everything presentational* merge the good
>   ideas from WA1, and start with the resulting draft.

This would be a huge waste of time and effort compared to going  
straight with HTML5.

-- 
Henri Sivonen
hsivonen@iki.fi
http://hsivonen.iki.fi/
Received on Wednesday, 25 April 2007 08:50:01 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 27 March 2012 18:16:09 GMT