Re: Rethinking @src in the context of chaining rules

Hi Manu

Yet another Manu test series:-) Thanks for these!

(B.t.w.: shouldn't all these cases end up in the test series? I think 
they should, because they reveal exactly those edge cases that might 
shake an implementation.)

Manu Sporny wrote:
> Mark Birbeck wrote:
> 
> That assumes that the 'current item' is something other than the image,
> and is defined elsewhere, like so:
> 
> ------------------------ BIM Approach -------------------------------

:-) I like the name:-)

> <span about="#current-item" rel="foaf:img">
>    ...
>    <img src="current-item.jpg" />
>    ...
> </span>
> ------------------------ Mark Approach -------------------------------
> <span about="#current-item">
>    ...
>    <img rel="foaf:img" src="current-item.jpg" />
>    ...
> </span>
> ----------
> <#current-item> foaf:img <current-item.jpg> .
> -----------------------------------------------------------------------
> 

To be precise: as far as I can see the first would work with Mark's 
approach, too.


> If an image is floating on a page, by itself, then we can't link it like
> we did above:
> 
> ------------------------ BIM Approach -------------------------------
> <img src="current-item.jpg" rev="foaf:img" resource="#current-item"/>
> ------------------------ Mark Approach -------------------------------
> <img about="#current-item" rel="foaf:img" src="current-item.jpg" />
> ----------
> <#current-item> foaf:img <current-item.jpg> .
> -----------------------------------------------------------------------
> 

That is correct.

> or you could do like it like this (Note how nice Mark's markup looks,
> the BIM markup looks wierd):
> 
> ------------------------ BIM Approach -------------------------------
> <span about="#current-item">
>    <span property="dc:title">A Picture of Me</span>
>    <span property="dc:name">Me</span>
>    <img src="current-item.jpg" rev="foaf:img" resource="#current-item"/>
> </span>
> ------------------------ Mark Approach -------------------------------
> <span about="#current-item">
>    <span property="dc:title">A Picture of Me</span>
>    <span property="dc:name">Me</span>
>    <img rel="foaf:img" src="current-item.jpg" />
> </span>
> ----------
> <#current-item> dc:title "A Picture of Me" .
> <#current-item> dc:name "Me" .
> <#current-item> foaf:img <current-item.jpg> .
> -----------------------------------------------------------------------
> 

It is correct and I _do_ agree that the BIM approach looks a bit weird. 
Having said that (and I said that before): for _any_ RDFa constructions 
there will be use cases where, well, RDFa will look weird (witness the 
problems raised by DanBri in another thread). We cannot win all battles:-)

> The important one, though, is how we apply rdf:type using @instanceof:
> 

Yeah, that is the one that started a whole thread of discussions...

> ------------------------ BIM Approach -------------------------------
> <span about="#current-item">
>    <span property="dc:title">A Picture of Me</span>
>    <span property="dc:name">Me</span>
>    <img src="current-item.jpg" instanceof="foaf:Image" rev="foaf:img"
>         resource="#current-item"/>
> </span>
> ------------------------ Mark Approach -------------------------------
> <span about="#current-item">
>    <span property="dc:title">A Picture of Me</span>
>    <span property="dc:name">Me</span>
>    <img rel="foaf:img" instanceof="foaf:Image" src="current-item.jpg" />
> </span>
> ----------
> <#current-item> dc:title "A Picture of Me" .
> <#current-item> dc:name "Me" .
> <#current-item> foaf:img <current-item.jpg> .
> <current-item.jpg> rdf:type foaf:Image .
> -----------------------------------------------------------------------
> 

Well... I believe that is not the case in Mark's approach. There are two 
cases: either we look at

http://www.w3.org/MarkUp/2008/ED-rdfa-syntax-20080103/

as it stands now, or include a change I actually proposed in

http://www.w3.org/mid/477E148C.3090009@w3.org

Let us analyse them separately.

1. In the _current_, documented approach the [new subject] in <img> will 
be set to a new BNode by virtue of @instanceof, which has a higher 
priority than @src. Ie, we would get a

[] a foaf:Image;
    foaf:img <current-item.jpg> .

2. However, I think Mark also agreed with my analysis that @src should 
move up the hierarchy. So look at

http://www.w3.org/MarkUp/2008/ED-rdfa-syntax-20080103/

with the difference that in step 2/3 @src handling must appear right 
after @about. That  means that [new subject] will be set to 
<current-item.jpg>. That will then be used for typing, ie, we will 
indeed get

<current-item.jpg> a foaf:Image .

However,  because @src is also kept to play a similar role to 
@resource/@href (in step 5 of the processing rules), @src is also used 
to establish the value of [current object resource], we will also get 
the weird triple:

<current-item.jpg> foaf:img <current-item.jpg>.

Unless I misread the syntax document, that case is definitely problematic...

A way to amend that in Mark's model would be the extra rule somewhere 
which says that if @resource, @href, or @src is used to set the value of 
[new subject] then it must be taken of the equation. But, well, what 
this would mean is that <img> will end up with a hanging rel, namely

<current-item.jpg> foaf:img ??? .

but this hanging rel will be never resolved, because <img> does not have 
any children.

Indeed, the BIM version, though a bit weird due to the presence of 
@resource, works, because @src is taken out of the process in setting 
the [current object resource] (not in the current syntax document but, I 
believe, the way it should be in the BIM model).

Ivan

P.S. (I _hate_ to say that, but this was always my use case on 
introducing @trel and @trev instead of a single @instanceof:-(. It is 
indeed the presence and the behaviour of @instanceof that messes up 
things here. However, the current BIM model with the @src behaviour 
seems to be a fair compromise after all...)

> Does that seem to be everybody's understanding of the differences
> between BIM/Mark with regards to @src?
> 
> -- manu
> 

-- 

Ivan Herman, W3C Semantic Web Activity Lead
Home: http://www.w3.org/People/Ivan/
PGP Key: http://www.ivan-herman.net/pgpkey.html
FOAF: http://www.ivan-herman.net/foaf.rdf