Tim Bray wrote:
On Jul 30, 2004, at 7:59 AM, Sam Ruby wrote:
I think there's a lot of value in encouraging reflection.
There's a lot of value in feed autodiscovery, thus the discussion around MarkP's RFC. But the decision as to which feed is autodiscovered out of some HTML is entirely up to the author of the HTML. You would be silly not to use autodiscovery if you were able and you had a feed you wanted to promote. But you have no obligation ("SHOULD") whatever. -Tim
...
I'm not sure how best to solve this problem, but putting a SHOULD in the document is the best solution I've seen to date.
OK, but someone has to come up with language to make it clear that this is a recommended good practice only in the case where the publisher wants to direct traffic to the feed. I am quite convinced that there will be lots of parasitic, "third voice", or "opposing viewpoints" feeds out there in the world, and the SHOULD clearly doesn't apply to them.
Just thinking it through out loud.
Person X is controversial. Person Y presents an alternative view. X.atom has a <link rel="alternate" href="X.html"> in atom:head Y.atom has a <link rel="alternate" href="Y.html"> in atom:head Y.atom has a <a href="X.html"> in atom:content
What should the recommendations be?
X.html should have a <link rel="alternate" href="X.atom"> in html:head Y.html should have a <link rel="alternate" href="Y.atom"> in html:head
Nothing more. In particular, X has no obligation, however slight, to include any reference to Y.
If we can agree to that general concept, we can work on the wording. The pace said:
the atom:feed element with type="alternate" points to an HTML page
Well, atom:feed elements don't have a type attribute. And "points to" is a bit vague. In the current draft, we are specifically talking about the URI contained in the atom:link element defined in section 4.2.2 where the type is "text/html" or "application/xhtml+xml" and the rel="alternate". The resource that that URI identifies SHOULD contain an autodiscovery link specifying this feed.