AFAICT, this has nothing to do with Atom-the-Protocol but only with Atom-the-Syntax, so CC'ing atom-syntax (and please follow-up there).
On Mon, Feb 16, 2009 at 9:25 AM, Erik Wilde wrote: > > hello again. > > RTFM, i guess. atom:source is expected to contain the *feed*'s metadata, not > the *entry*'s metadata from the originating feed. this almost solves my > problem, the exception being the question of how to identify the feeds that > are aggregated in a feed, and how to relate the entries to these aggregated > feeds. the <link id="" rel="via"/> and <entry via:source="" /> scheme might > still be one solution for that, allowing consumers to learn more about *all* > aggregated feeds, while pure atom clients could use atom:source to at least > understand where each individual entry came from. according to the extension > rules, via:source would have to be an element rather than an attribute, but > then it would be ok, i guess. and maybe xml:id for the link/@id? but it > seems like atom does not allow link/@xml:id, but any local attribute (but > only if a future version of atom actually defined it). My reading is backwards: undefinedAttribute in the RNC explicitly prohibits local:* attributes while allowing xml:id (and any "foreign" attribute besides xml:base and xml:lang) > so if link/@id and > link/@xml:id both are not allowed by the current spec, what other method > would there be to make an association between entries and via-links? <feed xmlns="http://www.w3.org/2005/Atom";> <link rel="via" href="http://example.com/feed-A"; /> ... <entry> <source> <link rel="self" href="http://example.com/feed-A"; /> ... And relate entries to li...@rel='via'] using the @href value. -- Thomas Broyer
