[whatwg] Inferring rel="feed" from the media type

On Wed, 29 Nov 2006, Mark Baker wrote:
> On 11/29/06, Ian Hickson <ian at hixie.ch> wrote:
> > On Wed, 29 Nov 2006, Mark Baker wrote:
> > >
> > > When you're documenting age-old practice which is in widespread use, 
> > > I fully agree.  Feed autodiscovery is effectively brand new and not 
> > > widespread at all when compared to how widespread it should become 
> > > in 20 years.  I think there's still lots of time to fix it.
> > 
> > I'm not sure what you're basing your assertion on; based on my own 
> > research of several billion documents, feed autodiscovery is used on 
> > hundreds of millions of pages, far beyond the point of no return in 
> > terms of backwards-compatibility constraints.
> 
> I wouldn't call that a very good metric for the purposes of this 
> discussion though, because I expect that the bulk of those pages are 
> produced by a handful of blog hosting services.  If we can shrink "100s 
> of millions" by 4 or 5 (or more) orders of magnitude with a handful of 
> persuasively written emails, then the situation is not what I would call 
> "widespread".

It's widespread _today_, such that UAs today can't change their behaviour. 
Thus we can't change the spec today.

If you reduced the volume of such usage, then it would be worth 
revisiting, but unless that happens, we're merely talking hypotheticals.

Personally I wouldn't be optimistic about the ability to change the legacy 
data; historically it has not been possible. I don't really know of any 
successful attempt, to the point where browsers historically have even 
tried using different processing modes -- the whole quirks mode thing -- 
to get around legacy content incompatible with the specifications.


> Are you able to analyze what proportion of those pages are hosted by the 
> top, say, 10 hosters?

Not from my current data set, no.

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

Received on Wednesday, 29 November 2006 19:52:23 UTC