This is one of the posts that has been backlogged for a while — I’ve just now gotten around to finishing it up. Sorry for the complete lack of timeliness on my part.
A feed autodiscovery link is a type of HTML
link element which tells web browsers and other
tools the location of a site’s syndication feed.
Deploying feed autodiscovery links for Atom and RSS has become
a widespread practice over the past several years, despite never
having been formally specified in a widely-recognized standard.
<link rel="alternate" type="application/atom+xml" href="/feed" />
This code snippet claims that /feed is an Atom document which is an alternative representation of the web page which contains the link.
There's a semantic problem with this use of
substitute versions for the document in which the link
occurs, and HTML
5 concurs, stating that
Yet on, say, a blog entry’s permalink page, you want to be
able to point at the blog’s feed without claiming that the
feed is an alternate of the entry — it’s an
alternate of the blog itself.
alternate keyword indicates that the referenced
document is an alternate representation of the current
What are the goals of standardizing feed autodiscovery?
- Semantic cleanliness
In November there arose not one, but four efforts to standardize this widespread practice:
The WHATWG’s Web Applications 1.0 (HTML 5) specifies a semantically-nicer
feedlink type, while grandfathering legacy markup like so:
alternatekeyword is used with the
typeattribute set to the value
application/rss+xmlor the value
application/atom+xml, then the user agent must treat the link as it would if it had the
feedkeyword specified as well.
This means that, in HTML 5,
<link rel="feed" href="/feed" />would be considered equivalent to my first example.
Here's a nice WHATWG blog post on it.
The RSS Advisory Board approved an RSS Autodiscovery document, which describes the currently-deployed feed autodiscovery mechanism.
Something about how the spec describes current "best practice", FSVO best that means "assume everyone else is an idiot"
Note that this spec is licensed under CC BY-SA 2.0.
Several months earlier, Mark Pilgrim prepared an Atom autodiscovery I-D for the Atompub Working Group. This draft was resurrected by James Snell as draft-snell-atompub-autodiscovery-00.
Like the RSS AB document, this draft set out to describe current practice. It immediately met some resistance on the ground of the misuse of
rel="feed"instead was proposed in PaceDifferentRelValue, though notably WA1's grandfathering language wasn't emulated in the pace.
In a series of paces (PaceRecommendFullURIsForAutodiscovery, PaceRestrictRelValuesForAutodiscovery, and PaceRestrictTypeValuesForAutodiscovery), Rogers Cadenhead proposed to bring this I-D closer in line with the RSS AB document.
These proposals directly tie into the "writing specs for idiots who didn't implement the other relevant specs" philosophy of the RSS AB.
Letting the I-D expire:
Rob Sayre something the RSS AB’s work, and blah feedautodiscovery.org