Re: embed RDFa --> embed coolness into Yahoo search results

On Tue, 17 Mar 2009, Ben Adida wrote:
> Ian Hickson wrote:
> > Your original point still stands, though.
> 
> No, it doesn't, please stop spreading these exaggerated points. Bugs are 
> bugs, and we'll fix them.

I think when the bugs include things like treating namespace prefix values 
as meaningful there is an indication of a pretty fundamental problem. It's 
not like that kind of bug just happens; there is no way that someone who 
understands the technology can introduce that bug, since there would never 
be any need for the prefix to ever be in the code in the first place.


> > More worryingly, though, I have to say that in trying to write these 
> > tests I had an extremely confusing experience reading the RDFa 
> > specification.
> 
> Did you read the spec or the primer? Which parts were confusing? The 
> questions you ask below are very easily answered by reading the Primer.
> 
> http://www.w3.org/TR/xhtml-rdfa-primer/

As far as I can tell the primer doesn't answer any of these questions. It 
just walks one through the process of doing this without explaining the 
edge cases.


> > When can one use href="" and when can one use resource=""?
> 
> Anywhere. @resource is meant to override @href when the clickable link 
> is not quite the same as the semantic link. This is useful only in some 
> cases, so I would agree that Yahoo could probably just use @href most of 
> the time. I'll mention it to them.
> 
> > Does using content="" with an absolute URL work also?
> 
> No. content is a literal. There used to be an example in the Primer of 
> this, but I think I must have removed them in the simpler rewrite. 
> That's unfortunate, and my mistake.
> 
> The idea is this:
> 
> <span property="dc:date" content="2009-03-17">today</span>
> 
> The point is to override the human-rendered when needed.
> 
> > What are the implications of using property="" instead of rel=""?
> 
> @property is for literal objects, @rel is for URI objects.
> 
> <span property="dc:title">The Title of the Book</span>
> 
> <a rel="dc:creator" href="http://ben.adida.net">The Author of the 
> Book</a>.
> 
> This should be clear from the Primer, Section 2.2:
> 
> http://www.w3.org/TR/xhtml-rdfa-primer/#id84624
> 
> > When does nesting matter and when does it not matter?
> 
> Nesting matters if you leave some "dangling triples" that can be 
> completed by nested statements. That's a more advanced topic, though, we 
> should probably make sure the above points are clear before we tackle 
> this one.
> 
> > Does it matter whan URL the assertions are made about, or will the 
> > SearchMonkey tool simply grab all the assertions from the document 
> > regardless of what URL they are about?
> 
> I think they grab everything but they link it to the appropriate URL. 
> Certainly it would be a bug if they merged statements about different 
> URLs. I'll check out your tests to see if there's one that highlights 
> this.

If you think my dad will understand the above answers then you are wildly 
more optimistic than I am.


> > What parts are necessary and what parts are optional?
> 
> Can you be more precise? In RDFa? In the SearchMonkey vocabularies?

Either. From the point of view of authors there is no difference.


> Given the questions you ask above, and given that plenty of other folks 
> haven't been nearly this confused

Are you sure? Yahoo certainly seem this confused. Even their examples have 
errors (e.g. one gives details about video.swf but them embeds game.swf -- 
and their tool doesn't care, you can change it to talking about a URL on a 
totally different page and it'll still give the exact same result).


> (Digg, many small CC publishers, etc...)

So when Digg says:

<img src="/environment/Climate_change_makes_us_boiled_frogs_says_Prince_Charles/t.jpg" 
     ...
     rel="foaf:thumbnail" 
     resource="/environment/Climate_change_makes_us_boiled_frogs_says_Prince_Charles/t.jpg" 
   />

...it's not an error?

From:
   http://digg.com/environment/Climate_change_makes_us_boiled_frogs_says_Prince_Charles

(My guess is that they're trying to set the foaf:thumbnail of the whole 
page, but my reading of the RDFa spec is that this sets the triple 
src-rel-resource, which is rather different, and in this case, rather 
pointless.)

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

Received on Tuesday, 17 March 2009 20:09:29 UTC