[microformats-discuss] URIs please!

[Peter Janes]

Kevin Marks wrote:
> On Jul 14, 2005, at 5:42 PM, Peter Janes wrote:
> 
>> Ryan King wrote:
>>> On Jul 14, 2005, at 8:14 AM, Bud Gibson wrote:
>>>> 2.  The presence of an XMDP cannot practically be assumed to mean  
>>>> that any occurrence of an attribute value defined in the XMDP is to  
>>>> be interpreted according to the XMDP's definition.
>>> Right, but we're not looking for individual attribute values, we're  
>>> looking for attribute values in context.
>>
>> I don't think it's necessarily true that a microformat will define a 
>> hierarchy of attribute values.  Take rel="nofollow"[1], which defines 
>> a single value, as an example: there's no structure involved, it is 
>> significant no matter where it appears in a document.  My proposed 
>> robots-exclusion profile[2] is an even more general case than 
>> rel="nofollow" since it defines attribute values (class names) that 
>> can be used on any subset of elements in a document in any combination 
>> or order.
[...]
> Right, Peter, but there is a difference between an elemental format , 
> which defines a single value, and a compound format like hCard.

As far as I can tell the main difference is what the context is.  For 
elemental formats the context is the entire document; for compound documents 
it's the content of all elements that match a top-level element in the XMDP 
hierarchy.

There's nothing stopping an elemental format--where all occurrences of its 
attribute values on a page are significant--from having an XMDP that defines 
its meaning.  (In fact, it's encouraged that all microformats have XMDPs, 
whether they're elemental or not.)  That means that elemental formats are the 
counterexample to the statement that, on a page referencing an XMDP profile, 
individual attribute values are meaningless unless they're contained by some 
other particular attribute value (which is how I read "we're not looking for 
individual attribute values, we're looking for attribute values in context").

I'd propose a new version of Bud Gibson's statement 2, on what linking to an 
XMDP means:

The presence of a link to an XMDP means that any occurrence of a *top-level* 
attribute value defined in the XMDP is to be interpreted according to the 
XMDP's definition.  Within that context, any occurrence of an attribute value 
defined below that top-level attribute in the XMDP's hierarchy is also to be 
interpreted according to the XMDP's definition.

> Validating the existence of an elemental format is easy, as it is a 
> single value; validating a more complex structure requires parsing.
> 
> http://www.microformats.org/wiki/elemental-microformat

I'm not sure "elemental" means "defines a single value".  On that page only 
the Rel* microformats define single values; VoteLinks, XFN and XOXO define 
many values yet are also considered to be elemental.  The distinction seems to 
be more one of structure: elemental formats define values that are applied to 
an attribute of a single element, while compound formats define values that 
are applied to the attributes in a hierarchy of elements.  (robots-exclusion, 
then, would be an elemental format.)

> Also, note the composability of the elements; you can make sense of 
> rel="tag" within another format, even if you don't necessarily parse 
> that format fully.

Unless, for some reason, another XMDP is being used that defines rel="tag" to 
have a different meaning.  One would hope that microformats.org will be able 
to avoid any internal overlap, but there's sure to be some group with their 
own format that redefines an attribute value to something more specific to 
their purpose.  Suppose the IETF defined a "references microformat" for RFCs 
that used rel="tag" to refer to documents created by the W3C's Technical 
Architecture Group; now your tag parser's going to eat a lot of bogus data 
because it presumes it knows what rel="tag" means.  Which gets back to the 
request that started this thread: to explicitly identify each microformat used 
in a document by its URI.

Sorry if this has rambled somewhat between topics; while writing I've been 
skipping through the list and various websites looking for references and 
examples to cite.  Feel free to divide and conquer as necessary. :)

Peter J.