Re: "Pave The Cowpaths" Design Principle

Matthew Raymond wrote:
>    I've pretty much come to the conclusion that any pre-defined
> attribute values that apply broadly to all elements (such as roles and
> class names) should be excluded from the specification. I was willing to
> make an exception for some class names if the definitions of their
> semantics and use were confined to something like metadata collection,
> but it's becoming apparent that there may not be sufficient use case for
> that.

Well, they are gone now :-)

>    Not having predefined values isn't sufficient to bar new predefined
> values. What differentiates |rel| is that there is a far more specific
> semantic context and an expectation that all values could potentially
> trigger processing by a user agent.

OK, so the second half of my point, then.

>> Also, "rel=nofollow" almost seems too simple to be a microformat.
> 
>    What exactly is your definition of "micro"? Should we call it a
> "nanoformat"?

This is the question I'm asking. Is a microformat just "some stuff some 
unknown number of people have agreed to do to add some level of extra 
semantics to web pages"? Or is there more to it?

If you and I agree that "class='gerv'" should be used to indicate 
content about me, does that make it a microformat?

>> Is  "autocomplete=off" a microformat too?
> 
>    No, because it's an attribute and thus causes the document not to
> validate. It's really a de facto standard, actually.

Er, rel="nofollow" is an attribute.

>> We seem to be stretching the 
>> definition of "microformat" quite a long way, such that it is losing 
>> meaning.
> 
>    I don't think excluding |rel|-based extensions of HTML from the
> definition of "microformat" serves any useful purpose.

That's not what I'm saying. I'm saying "if that's not excluded, what is?".

Gerv

Received on Thursday, 17 May 2007 11:53:07 UTC