Re: [standardizedFieldValues-51] new issue on deploying new terms, media types, properties, elements, etc.

Dan Connolly wrote:
> * Atom seems to have gotten it "right" lately...
> 
> [[
> If a name is given,
>    implementations MUST consider the link relation type to be equivalent
>    to the same name registered within the IANA Registry of Link
>    Relations Section 7, and thus the IRI that would be obtained by
>    appending the value of the rel attribute to the string
>    "http://www.iana.org/assignments/relation/".
> ]]
>  -- http://xml.coverpages.org/draft-ietf-atompub-format-08.txt
> 
> hmm... or did it? "Note that use of
> a relative reference is not allowed."

This rule exists because, in Atom, xml:base is always in play.

------------
...
<entry xml:base="http://example.org/bar/">...
   <link href="../foo" rel="foo">...
</entry>
...
------------

The WG saw no reason to allow stuff like "../relation/via" as a relation 
value, because that would mislead implementors who don't give the spec 
an extremely close read. @href in the example above would resolve to 
"http://example.org/foo".

"The value of 'rel' MUST be a string that is non-empty, and matches the 
'isegment-nz-nc' or 'IRI' production in [RFC3987]."

So, a subset of the possible relative references are allowed, and they 
are always resolved with simple string concatenation. Hope that makes sense.

Robert Sayre

Received on Thursday, 12 May 2005 22:14:56 UTC