[whatwg] Joe Clark's Criticisms of the WHATWG and HTML 5

>> * metaphorical semantics (the meaning of the things)
>> * structural semantic (the order of the things)
>> * purposal semantics (the purpose of the things)
>>
>> Let me denote some examples:
>>
>> * metaphorical:
>>   - <joke> marks something to be funny, sarcastic or zynic
>>   - <strong> emphasises a content strongly
>> * structural (aka. define dependencies):
>>   - <p> marks text to belong together, to be connected sentences
>>   - <chapter> marks sections to belong together
>>   - <formular> marks an interactive region
>> * purposal:
>>   - <control> marks a region to be inside an interface-element
>>   - <nav> marks a section to contain the/a navigation
>>   - <title> marks a region to be a chapters/documents title
>
> It's vaguely similar to some classification I was trying to do not long
> ago, although I could not come to something conclusive.
>
> Where would you put <a>, <img>, and <i> in this?

 There are some concepts/functionalities hidden in tags that you
can't map into CSS or JS. While you may create the <a> by a
<div>+<style>+<script>, you may not do so elegant with <table>.
 This is talking about equivalent realization of actual behaviour
& presentation, stripped of all semantics.

 That's what I really meant/mean: breaking semantics by no means
breaks functionality and appearance. To preserve that or to
support that further, it's nessessary to abstract and split
semantic, behaviour and presentation completely.

 There is no conceptual difference between <a> and <address>, so
both are purposal, they are use-less information without
communication or action. I may consume a poem, but not street-
names, or an index - I can't _use_ a poem (well ...), but I can
act upon what I read in an index, or an address. You can stress
generalization by using <address type="anchor">.
 And maybe against the stream: I think it should be
<a alt="text(something)">http://url</a> if there would not be the
need to put HTML instead of 'something' - <a alt="element(id)">
comes to mind. 'url' is the original information, not auxiliar
and 'something' an alternative one.

 The advantage of textual information is, that it's easy
transformable into visible or aural information, that is
(actually) not the case with visual or aural information (they
can not easily be converted into another form of information).

 So <img>, <sound>, ... will always be a form of presentation,
and not description. So they should be categorized as alternative
*prepared* presentation, like <p alt="url(pics/example.png)">.
Which allways can be overwritten by automatic presentation. A
possible structural tag <graph> (which could be inherited from
<table>):

<graph alt="url(pics/graph.png)">
	<axis><x>val1</x><y>val2</y></axis>
	<stamp><x>1.0</x><y>2.5</y></stamp>
	<stamp><x>1.4</x><y>2.7</y></stamp>
</graph>

 is machine-processable and semantic, can serve the superduper
made PNG or an alternative presentation.

 <i> has no semantic meaning per-se, too much different concepts
utilize italic presentation that it's significance is a wild guess.

> Michel Fortin
> michel.fortin at michelf.com
> http://www.michelf.com/

 Ciao
	Niels

Received on Wednesday, 8 November 2006 09:51:07 UTC