Re: [draft-nottingham-http-link-header-06] 4.1 Registered Relation Types

On 24/07/2009, at 11:55 PM, Anne van Kesteren wrote:

> I think this section should not talk about what implementations are  
> allowed to do internally. The concern of the specification should be  
> what goes in and what goes out. Therefore I think the entire last  
> paragraph should be dropped. We should not give the impression that 
>  will _ever_ work.

I was sorely tempted to do this a little while ago. Unless someone  
objects strongly, I'm going to drop that text (currently a separate  

> Also, can we require that the canonical registration form is  
> lowercase?

That's probably sensible. I'll modify the BNF and prose.

Additionally, any objections to requiring sgml-name (probably needs a  
new name, now that it's going to be all lower) for registered?  
Currently it's SHOULD.


Mark Nottingham

Received on Friday, 21 August 2009 06:41:55 UTC