[microformats-discuss] Extending the embedded information

[Ryan King]

On Jul 8, 2005, at 9:25 AM, Tantek Çelik wrote:
> On 7/1/05 3:03 PM, "Carl Beeth" <carl.beeth at gmail.com> wrote:
>>
>> ...
>>
>> The reason the class is important in on the link is that it allows  
>> the
>> user agent/crawler to identify what is behind the link. a possible
>> alternative would be to use type="text/vcard" but sadly I think they
>> are supposed to be mimetypes but could not find the mime type for
>> vcard
>
> One of the key prerequisites for proposing a microformat is to  
> first exhaust
> the possibilities of using 1. a semantic XHTML element (or  
> attribute), and
> 2. a semantic XHTML compound.

See http://microformats.org/wiki/process for some more details on  
this sort of stuff.

Also, if that document doesn't help/needs improvment, feel free to  
complain. :-)

> In otherwords, if you can do it with semantic XHTML, you don't need  
> (and
> shouldn't propose) a microformat.  Re-use before inventing.
>
> In this case you are trying to develop a mechanism for linking to an
> "alternate" version of something.
>
> This is precisely what rel="alternate" (defined in HTML4) does.
>
> As far as claiming what is at the other end of a link, that is not  
> something
> that you can reliably do from the source of a link.  Heck, the  
> destination
> of the link might not even exist.  This is fundamental to the web.   
> All you
> can do is provide a hint.

To add to this... trying to assert something precise about something  
on the other end of a url is a form of tight coupling that doesn't  
work well on the Web in general. Its possible that the assertion will  
be wrong at some point in the future. Its better to just not try and  
make a strong assertion and only provide a hint.

> Now you might say that rel="alternate" is insufficient because it  
> doesn't
> distinguish between an "abstract" and a "full" version, or, as  
> others have
> found in other contexts (e.g. localization/translation), the  
> "original"
> version, and a "copy" or a "translation".
>
> That's a discussion worth having, as I think there is enough of a  
> need there
> to propose an elemental microformat to address one or more of those  
> use
> cases.  However, I do think that a single rel value might suffice  
> for this,
> rather than a class name for each microformat.  The latter seems  
> like an
> unnecessary explosion in the number of class names.

Two thoughts, elaboration again:
1. This kinda universal microformat ideas (things that will or could  
apply to all microformats) need to be approached very carefully.
2. Let's remember that we want to the most meaningful xhtml we can.  
In this case, rel is more meaningful than class. Generic elements and  
attributes like <div>, <span> and class= are somewhat of last resorts.

> For now though, just use rel="alternate" to link to a page that  
> contains
> another version of your hCard, perhaps a more complete version.   
> That should
> actually be sufficient for a user agent/crawler to go get the complete
> version.