Richard Salz wrote:
There is no "for" attribute in RFC 4287. If this use case requires it,
maybe a link relation isn't the best way to do it (instead an extension
attribute?).
Can you show me what you mean by extension attribute? It might not be
common, but adding attributes to link elements seems legal according to
the RFC.
I would assume that they'd need to be in a namespace different from the
default namespace, otherwise we would have a potential collision with
future revisions of RFC 4287.
The QName can be avoided by using the "Clark notation"
({namespace-uri}local-name).
Is there a common XSD datatype for this? I agree that if that's the
strongest argument I can come up with, then... I don't care much one way
or the other, whatever the group prefers.
I personally do not have problems with QNames in attribute values (IMHO
that ship has sailed); but I wouldn't be surprised if others disagree.
In this case it could easily be avoided.
It seems that collides a bit with RFC 4287's definition of type:
I thought it was clearly in the spirit -- you can get whichever type of
schema you want. Can you be a bit more specific for me?
The type is only a hint; thus a SHOULD-level requirement not to have
multiple instances of the same type seems to conflict with the advisory
nature of the attribute...
> ...
BR, Julian
