Re: Report on testing of the link relations registry

On Mon, 16 Aug 2010, David Singer wrote:
> On Aug 15, 2010, at 3:02 , Ian Hickson wrote:
> > 
> > I tried Googling "RFC-nottingham-http-link-header-10". This led me to:
> were you not able to follow the link on the IANA page?
> search for "link relations" on
> Or is googling preferred for those that work at google :-)?

I doubt most people who are interested in registering a link relation 
would think of looking on It didn't 
actually occur to me to try -- even looking at it now, I just assumed it 
would link to the registry (which I'd already found) and not the 
specification about it.

> >
> > 
> > This is a 27 page document. A quick glance at the table of contents 
> > was unhelpful, but a scan through the document revealed section 
> > "6.2.1. Registering new Link Relation Types", which for some reason is 
> > not in the table of contents.
> The table of contents cover sections and their first-level sub-sections, 
> pretty clearly.  Section 6.2 has a pretty obvious title, in the index.
> "6.2.  Link Relation Type Registry . . . . . . . . . . . . . . . 10"

My experience with previous RFCs had led me to believe that this was the 
section describing the creation of the registry, not the use of the 
registry (which is indeed the case, though there is a subsection, not 
listed in the table of contents, on using the registry).

> The overall these of this email seems to be that it's easier to edit a 
> wiki, than to write a stable specification, and register the code-points 
> in it in a formal registry.  Did we really need such an email to know 
> that, or am I missing your point?

I expected registering the link relations to be somewhat tedious. I didn't 
expect to actually fail to register everything I tried to register, 
especially given that the people running the registry are aware that my 
goal was to find out how easy it was to register something. It's not a 
matter of a wiki being easier -- as you say, that's pretty clear. It's 
more about whether the additional complexity is minimal, and acceptable, 
or whether it is so much additional complexity that interoperability will 
suffer as a result. Given how hard it is to register link relations, I 
would expect this registry to quickly become as useless as the MIME type 
registry, and essentially be ignored by the overall population. That isn't 
helpful to the Web's long-term health -- we'll just end up with people 
doing what they do now, making up link relations without even checking to 
see if they might clash with someone else's, without conformance checkers 
being able to check what relations are ok and which are not, etc.

Ian Hickson               U+1047E                )\._.,--....,'``.    fL       U+263A                /,   _.. \   _\  ;`._ ,.
Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'

Received on Monday, 16 August 2010 18:02:11 UTC