I'm the chairman of the RSS Advisory Board, which has published our first autodiscovery specification [1]. I'd like to participate in the drafting of Atom's effort in this area with the goal of making it possible for publishers to support autodiscovery in the same manner regardless of syndication format.
Our original spec said that it could be used with RSS 1.0, RSS 2.0 and Atom, but the Atom guidance was removed to get out of your way as your spec is being drafted. I will put a couple of proposals on the wiki this morning. Regarding your current draft, a couple of editorial suggestions: 1. Are the "most relevant rules" part of Sections 3.2 and "most relevant differences" part of 3.3 necessary? It's helpful information, but it documents behavior that's covered by the HTML and XHTML specs, so it seems redundant and makes your spec longer than it needs to be. 2. The href attribute's section is out of order alphabetically. For easier reference, I'd order sections 4.1 through 4.3 as href, rel and type rather than rel, type, href. 3. Your introduction to autodiscovery doesn't describe the most common and popular implementation of the technique. Your graf: Autodiscovered Atom feeds may be presented to the user in a variety of other ways. In the past, Atom-enabled clients have implemented local proxies that monitor visited web sites and notify the end user of autodiscovered Atom feeds in real time. Such notification is also built directly into some desktop web browsers. My suggested rewrite: Autodiscovered Atom feeds can be presented to the user in a variety of other ways. Current versions of the Microsoft Internet Explorer and Mozilla Firefox browsers notify users of the presence of such a feed by displaying an orange feed icon in the browser's address bar. Clicking the icon initiates the process of subscribing to the feed. 1: http://www.rssboard.org/rss-autodiscovery
