* Becker, Matthew R <[EMAIL PROTECTED]> [2007-03-03 01:15]: > Can anyone elaborate some more on why draft 15 is changing the > standard ordering to be based on app:edited instead of > atom:updated? Wouldn't feed readers be triggering off of > atom:updated?
Aside from the fact that feed readers should not be polling the collection feed in the first place, they don’t expect any particular ordering anyway. How to order the feed is the publisher’s choice, who can decide what entries aggregators should have the best chance of staying up to date with. > Leaving that as the default sort order would allow for > optimization in determining significantly changed content as > opposed to ordering on app:edited where changes may or may not > have been significant. There is no default order. RFC 4287 says no particular meaning can be derived from the order of entries in a feed. AtomPP imposes a constraint on that. The rationale for ordering is to allow efficient synchronisation of local caches by AtomPP clients. If app:edited changes, atom:updated doesn’t necessarily. If atom:updated changes, however, app:edited will also. Therefore, if the collection feed was ordered by atom:updated, minor updates could be buried arbitrarily far back in the collection feed, and clients can that keep a local cache would have to page through the entire collection to synch with the server. In contrast, if it’s ordered by app:edited, all edits (including significant) will bubble to the start of the feed, and the client can stop paging as soon as it sees an entry that was last edited before the last client synch. Regards, -- Aristotle Pagaltzis // <http://plasmasturm.org/>
