Re: finding closure on the issues

Ben Adida wrote:
> 
>  - @resource/@href completing hanging @rel, and
>  - @src equivalent to @about or @resource/@href.

Manu speaks of complexity: I think there isn't much complexity
difference between the two models from *our* standpoint as developers of
RDFa. But from the *HTML author*'s standpoint, I think we trigger
significantly more complexity if we allow @href to complete @rel.

I believe we should

1) *not* have @resource/@href complete hanging @rel

and, *assuming we go that way*,

2) @src should remain equivalent to @resource/@href (not @about). Yes,
this is a change from my proposal, since, as I noted, it conflicts with
the reasoning for (1).

HOWEVER, if, as a group, we choose to have @resource/@href complete
hanging @rel, then I think the right answer for (2) is to have @src be
equivalent to @about.

===============

Here's my reasoning: I worry about what typical HTML authors will write,
and if what they write yields un-meant triples on a regular basis, that
will be bad for RDFa adoption.

Specifically:

- @href completing a hanging @rel already caused confusion with Creative
Commons examples developed with a hanging @rel (independently of me).
Links added for clickable reasons suddenly completed a hanging @rel, and
the outcome was non-sensical. This is a *real* example of folks who
understand RDFa relatively well already.

- if @href can complete a triple, then for consistency, @instanceof
*has* to apply to that @href. For example:

<div about="#me" rel="foaf:knows">
   <div instanceof="foaf:Person">
      <span property="foaf:name">Mark</span>
   </div>
</div>

which means

<#me> foaf:knows _:bn0 .
_:bn0 a foaf:Person .
_:bn0 foaf:name "Mark" .

Now, add @href on the inner DIV:

<div about="#me" rel="foaf:knows">
   <div href="#mark" instanceof="foaf:Person">
      <span property="foaf:name">Mark</span>
   </div>
</div>

If @instanceof doesn't apply to #mark, then we are breaking up the
chain, which is counter-intuitive.

*BUT*, what this now means is we're back to @instanceof shuffling back
and forth between @about and @href, which is exactly what folks
(rightfully) complained about with my previous chaining model.

In other words, @href completing @rel triggers significant complexity to
*HTML authors*.

===========

Without consideration for chaining, it's clear to me that @src is a lot
more like @about than it is like @href. In particular, if you have:

<img src="ben.jpg" instanceof="foaf:Image" />

You really mean that the image, <ben.jpg> is a foaf:Image.

Also, since @src means "inline-object," and @href means "remote object",
I foresee a future where:

<img src="thumbnail.jpg" rel="thumbnailOf" href="big.jpg" />

makes sense, and thus having @href override @src is wrong.


Now, if @href can't complete @rel, then for the same author-complexity
reasons, neither should @src, and thus @src cannot, for practical
reasons, become the equivalent of @about.

However, if we come to the conclusion that @href should complete @rel,
then it is also acceptable that @src complete @rel, and so the model of
@src being equivalent to @about makes a lot more sense.

So my vote on this issue depends on how we resolve the previous one: if
we opt to ignore author complications, then we should opt for the
cleaner model for @src.


-Ben

Received on Monday, 21 January 2008 19:38:53 UTC