Eric Scheid wrote: >On 6/5/05 1:07 PM, "Nikolas 'Atrus' Coukouma" <[EMAIL PROTECTED]> wrote: > > > >>Hrm. This is an interesting point. I'm not too concerned about "find >>every feed, regardless of relevance" because I think only search engines >>will be interested in it, especially if all the other cases are marked. >> >> > >finding every feed is not my concern either. > > > >>They can bear to check the feed and see what the root element is. >> >> > >this won't work ... see below. > > > >>This also makes rel="alternate" seem like an even worse choice for >>*feed* autodiscovery because it would make sense to link to an atom >>*entry* as rel="alternate" from the page for an individual entry. >> >> > >absolutely! > > > >>I really don't think @rel is the place to address concerns about type. >>That's really the job of @type (of course). If we need to declare more >>mime-types, then so be it. >> >> > >Just to throw more fuel on the fire: > >It is quite conceivable for an Atom Feed Document (AFD) to contain a set of >entries which won't grow or be updated, such as an AFD which contains all >postings for a calendar period, or an AFD which contains one entry for each >chapter of a book, and so on. > >Thus, neither mime-types nor root-element-sniffing will be reliable enough >to discover the resource which is appropriate for "subscribing" to - ie. >discovering which Atom Feed Document is the one which will be updated as >time goes by in the usual sliding window manner, and not the monthly archive >that page happens to be contained within. > > > >e. > This warrants groaning. I think it's safe to say that user agents shouldn't subscribe to entry docments. Feeds may be worth bothering with.
Source: (HTML/XHTML) Clearly the friendlier end to do autodiscovery on. Remember, the goal here is to indicate that the URI from @href is worth subscribing to. I'm running out of attributes to abuse. >From the XHTML Link Module: [1] charset, href, hreflang, media, rel, rev, type Link Module: [2] target All of these are listed in the 4.01 spec[3]. There's also the slim chance of using XLink's link semantics [4]. They seem to be mostly redundant with @target. Speaking of @target, I'm currently favoring it because it's used for indicating where a link whould be loaded. This makes a hair more sense with today's browser-feed integration. Destination: (Atom) The only way I can see to indicate that a feed is "alive" and worth subscribing to (assuming existing Atom spec) is to include cache control (Expires or Cache-Control headers). [1] http://www.w3.org/TR/2004/WD-xhtml-modularization-20040218/abstract_modules.html#s_targetmodule [2] http://www.w3.org/TR/2004/WD-xhtml-modularization-20040218/abstract_modules.html#s_linkmodule [3] http://www.w3.org/TR/html4/struct/links.html#h-12.3 [4] http://www.w3.org/TR/xlink/#link-semantics -Nikolas 'Atrus' Coukouma
