On Feb 4, 2008 8:57 AM, Daniel Aleksandersen < [EMAIL PROTECTED]> wrote:
> Why does so many in this list fail to eralise that I am talking about the > <feed> element's <id> element? Not an <entry>'s? I see these messages > were people think I am talking about entries all the time. That is wrong. > I know that they should uniquely identify the entries in the <entry> > element. But what about the the <feed> element? Hi Daniel, FWIW, in the GData implementation of AtomPub we keep the feed-level atom:id constant across all derived views of a collection/feed, and we support a wide variety of query models (by-date, pagination, by label, etc). Our rationale was consistent with what others have said on this thread, that it's beneficial because it can help to correlate that two different types of queries have the same origin. A related intepretation is the perspective that queries/pagination don't create a new logical feed, which giving them a unique id would imply. We do, however, provide a feed-level self link relation href that contains any query parameters, making it possible to retrieve the same (page, date-range, ...) logical view of the feed again. Thus, it "identifies a resource equivalent to the containing element". Cheers! -- Kyle
