[whatwg] foreign attributes Re: several messages about XML syntax and HTML5

[Ian Hickson]

On Tue, 5 Dec 2006, Elias Torres wrote:
> > 
> > > <div> <span class="ibm-part-description">our part number <span
> > > class="part-id">123</span></span></div>
> > 
> > Yes? What about it?
> 
> I guess this is similar to Karl's example.
> 
> <div id="order1" class="ibm-order">
>    <span class="ibm-part-description">our part number
>        <span class="part-id">123</span>
>    </span>
> </div>
> 
> part-id would according to microformats be a property of 
> ibm-part-description existential instead of order1. Is there way to 
> specify that kind of behavior. Would include do the trick here?

I don't understand what you mean by "existential" in this paragraph.

If you are saying "how do we make the pard-id be a property of the order 
instead of the description", then the answer is:

   <div id="order1" class="ibm-order">
    <span class="ibm-part-description">our part number</span>
    <span class="part-id">123</span>
   </div>


> > > Microformats is very restrictive in how you can parse the data.
> > 
> > I don't understand why you think this. Could you elaborate?
> 
> I guess restrictive is the wrong word, it's only defined to say you must 
> grab the entire HTML content within the element. Fair enough again. 
> Although, I'd like to differentiate between HTML sub-content, plain text 
> and also specify which data type. quantity is an integer/long/etc.

Well, you can define your structure pretty much as you like, nothing 
forces you to define "description" in the example above as being the 
entire contents, you could just as easily say that your processing model 
only involves the direct child text nodes, or some such.


> > > We need the flexibility of specify the content everywhere, yet the 
> > > property apply to any element on the page, not just the parent 
> > > element.
> > 
> > Again, concrete examples would really help me understand this?
> 
> In some part of the page....
>
> <div id="picture1">
> <img src="picture1.jpg">
> </div>
> 
> Now somewhere else in the page....
>
> <div>
> <span about="picture1" property="dc:title">A Night in Venice</span>
> <div>
> 
> Here I'm able to connect two elements in the page. I'm not sure how I 
> would use id to differentiate the two.

Why would you need this? Just have:

   <figure>
     <img src="picture1.jpg" alt="">
     <legend> A night in Venice </legend>
   </figure>

If you could provide a URI to a document showing why you need the legend 
and the figure split from each other, that would be very useful.


I can't emphasise enough that it would really help if you could point to 
some real world examples (with URIs) showing what we are talking about. 
Currently, as I've mentioned before, it feels very hypothetical.


On Tue, 5 Dec 2006, Elias Torres wrote:
> 
> I think we keep focusing on the ways to extract information and I don't 
> disagree that we could find a million ways to hack class attributes to 
> do so. I can see myself doing class="about-subjectName", 
> class="predicate-propertyName", and so on. If anything, what I'm asking 
> for is a less hacky way of using "ibm-xxxxx" for everything and maybe 
> have a way to denote a prefix and a property attribute to different 
> properties from classes. Let me know if this would bring down HTML5 if 
> we were to try.

I really don't follow. Why do you need prefixes? What is "hacky" about the 
well-defined and accessible extension model in HTML today?


> On the other hand, we keep missing the point, that no matter what the 
> syntax is, in microformats at least (our current answer) we can't 
> differentiate from class values that are properties/classes(types) and 
> which ones are not.

You know which are which. You invented them. (Although, what's a class 
value that _isn't_ a "property/class(type)" ?)


> In our scenarios that we have been exchanging is nothing but ideal to 
> see how the structure maps to our data and how easy is to write JS to 
> get at it. However, we are looking for a mechanism that focuses on 
> property attribute to know when someone indicated a property as opposed 
> to just a style.

Don't use class for styles. Problem solved.


> It's like imagining a generic microformat extractor that generates JSON 
> objects on any random webpage with a lot of different styles and a few 
> hCards here and there. Is this really ideal? I think that before HTML5 
> microformats did what it could with what is there (not to say that HTML 
> isn't semantic at all today), but if we can define more clearly defined 
> attributes to do so, why not?

Could you show me Web pages that are trying to do what you are referring 
to, which would benefit from what you are describing? It would really help 
me understand what problem you are trying to solve.


> Do you see any workable solution?

To be honest I don't understand the problem, let alone have a solution.

-- 
Ian Hickson               U+1047E                )\._.,--....,'``.    fL
http://ln.hixie.ch/       U+263A                /,   _.. \   _\  ;`._ ,.
Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'