Re: Report on testing of the link relations registry

On Tue, 17 Aug 2010, Tantek Ã~Gelik wrote:
> On Mon, Aug 16, 2010 at 4:25 PM, Silvia Pfeiffer 
> <silviapfeiffer1@gmail.com> wrote:
> > 
> > I believe if we set up another registry, we will eventually run into 
> > the same problems -
> 
> I believe this is not only an unfounded supposition, but there is 
> evidence to the contrary.
> 
> I.e. so far that hasn't happened.
> 
> Two successful counter-examples:
> 
> * http://microformats.org/wiki/ and in recent years the very thorough
> documentation of rel values here:
> http://microformats.org/wiki/existing-rel-values
> 
> * The WHATWG rel value registry
> 
> And even though they are not perfectly harmonized, so far, in
> practice, that hasn't been a problem, and over time, I expect that
> further harmonization will occur.

Note also that the WHATWG rel registry hasn't really been actively 
maintained yet, because the spec hasn't until recently been at all stable, 
and since becoming more stable the maintenance of the registry has been 
put on hald waiting to see what happens with the rel registry issue. If
we begin active maintenance I would expect it to quickly converge with the 
Microformats wiki page.


> Meanwhile I will humbly suggest that until a comparably easy 
> process/procedure is proposed, that we (the HTMLWG) stick with what is 
> working today, and has worked for many years, that is,
> 
> 1. permit any author to propose/experiment with new rel values on the 
> web (as HTML4 originally allowed, and http://gmpg.org/ successfully took 
> advantage of and deployed with XFN)
> 
> 2. continue maintaining microformats.org's registry of existing 
> rel-values (as that's been the source of the vast majority of successful 
> rel-values over the past 5+ years)

That would be fine by me too. We would need to add the two flags (how the 
relations affect <link> and <a>/<area>) to fully address the needs of the 
spec, but that is presumably fine.

My ideal would be something more like what Lachlan suggested, probably 
supported by a wiki for documentation purposes.


> 3. continue experimenting with the WHATWG wiki registry of rel-values
>
> 4. encourage those that have the patience to essentially document the 
> successes that come out of those aforementioned registries by creating 
> RFCs for them (i.e. as rel-license eventually made its way from a 
> microformat to an RFC).

Any spec should probably be fine, not just RFCs. In particular, for 
instance, the "pingback" relation has had an old and stable spec for 
years (it even has multiple translations); I see no benefit to recasting 
that as an RFC or in fact changing it in any way -- we should just 
reference the existing spec and be done with it.

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