Re: Report on testing of the link relations registry

On 21.08.2010 19:31, Tantek Çelik wrote:
> ...
> As far as what elements a particular rel value applies to, I suggest a
> single column: "element restrictions" with permissible values:
>
> * blank or empty - no restrictions, applies to<a>,<area>,<link>
> elements in HTML
> * "only a, area" - only applies to "a, area" elements for example.
> comma separated list of applicable elements.
> * "not link" - applies to everything but the HTML "link" element for
> example. comma separated list of elements that the rel value does NOT
> apply to.  Of any non-empty restriction, I see this one being used the
> most often.

Wait -- are you saying that there are link relations that do not make 
sense on <link>? I think that's totally backwards (I just sent a mail in 
a separate thread about this -- 
<http://lists.w3.org/Archives/Public/public-html/2010Aug/0225.html>).

> If this seems sufficient for the information needed for HTML5, I can
> go ahead and add this column to the table of existing rel-values.

Well, this would not address the "effect on link" aspect; and that was 
the example that was raised back last year as use case for "additional 
flags in the registry".

(Validation is interesting as well, but I think the question of "what 
can I derive from the presence of a certain link relation even if I 
don't know its precise semantics?" is much more important...).

Best regards, Julian