[uf-discuss] class="url"?

[Siegfried Gipp]

Am Sonntag, 5. November 2006 18:33 schrieb Chris Casciano:

> I'm still looking for the meaning of alternate as well in these
> cases, because i haven't quite figured out what its an alternate to/of.

just imagine several alternate contact infos in a vEvent or maybe several 
alternate postal addresses or several alternate email addresses of the same 
type (more than one home email address). O.k. several alternate names is a 
bit extreme... :)



> The question: if I simply wrapped each of the above cases in <address
> class="author vcard"></address> and linked the "obvious" links with
> class="url" (emails, personal web sites, etc) but nothing else would
> that be a "good" hcard semantically? How about if I also linked some
> of the other content such as "disclaimer" or "© 2006" or linked
> "London, England"or "Great Russell Street"  to google maps? should
> those links be attached to the contact information or simply be left
> with the raw content? What if I wanted to explain what a web
> developer was by linking to wikipedia? that's certainly fine by
> semantic hypertext standards, but its not contact information. So am
> I wrong for wrapping the entire line a container with class="vcard"
> or do we infact need some method of explicitly stating what <a href>s
> are (or are not) part of that contact information?

O.k., that means that the class name "url" does not mean any arbitrary url, 
since it means the one specific url. That's fine. But it is different than 
the w3c has defined url. For me that's fint, too, but it should be clarified. 

>
>
> And these are simple example of quite compact content. Look at cases
> like flickr or linked in profiles where there is quite a lot of data
> about a person being displayed but as a result the "hcard" specific
> elements are scattered about the entire page (or a large portion of
> it).. Those cases must have a way to separate the "hcard" specific
> email addresses, urls, etc from the many, many other links in the
> body of the containing element (FLICKR: friends profiles, group list,
> "change information", "testimonials" are all between the block with
> name and location and the other block with contact / email address)

O.k. let's assume the hCard data scattered around the whole page. Then, 
somewhere you would have a name. Let's assume Lucky Luke:
<span class="vcard fn">Lucky Luke</span>
Fine. Legal. But then, what would this mean:
<a class="vcard fn" href="http://www.wildwest.com">Lucky Luke</a>
The url is still very closely bundled with that name. So this simply implies 
that this url contained in the href attribute is the url for the person named 
in the content of that container. So why add a class name of "url"?  Any 
other information about that person may be scattered around the whole page, 
marked up in a similar way. In any cases it would make sense to add a url, 
you simply do the markup with the <a> element. Any other <a> element without 
hCard class names is then any other random url. It may or may not be related 
to this person. So if you like to add a google earth link, well, why not. 
There is no microformat for this. So you cannot make a statement about if 
that link is related to that person. Except if you enclose all that within a 
container, which then implies, that all content of that container is related 
to this person. That again would make it unnecessary to add class="url" to 
any links.

In the above two examples it's the same: Anything in the container is related 
to ist content. In the first example it's simply the name. Since <span> does 
not imply anything about its content (except of being inline content, not 
block content), it is reasonable to make an additional statement about its 
content. So with vcard fn this is done. The second example has one more 
information. Still all the content of the container is related to that 
person. This is true for the href attribute too. But the content type of the 
href attribute is already specified. The <a> element together with the href 
attribute do already have a well defined semantic. Contrary to the span 
container. 

I think it is better to use existing semantics and add only where necessary.

Besides that: If the information would be really scattered over the whole 
page, and worse: Mixed with data about other persons, then it would not be 
possible to automatically extract the various address records. You do not 
know which data belongs to which person. The only thing you can extract is 
the name with the url related to that name, if you use the <a> element as 
container. In this case this particular url is definitely related to exactly 
this person, no other. No need to add a class name of url.

Just an idea aside: If you scatter the information over the whole page, but 
the whole page is only about one person, it would be useful to apply the 
vcard property to the body element. Then any url in this page is related to 
this person. But not as close as related to the content of the <a> element. 
So if such a page would contain a google earth link with the street address 
in the <a> container, then this link would be more related to that street 
address than to the person. The most close relation with a person is with its 
name. In this case the link related to the name would be the link to be 
exported to the VCARD URL property. Still unambigeous, still no necessity of 
using some url class name.

O.k. ease of adoption might be an argument. And specifically pointing that url 
as something special might be an argument, too. But in both cases the meaning 
of the class name url and the url as defined ba the w3c is different.