On Monday, May 2, 2005, at 07:22 PM, Bob Wyman wrote:
Antone Roundy asked:Yes, at least normally. An entry should have only one "source."Category feeds: ... Should they, or feeds that combine category feeds present the entries like aggregated feeds?
On a site that has a "master" feed and also provides copies of
entries in a variety of category feeds, I feel that the "master" feed should
be the source and things in the category feeds should be copies which
contain atom:source elements pointing to the master. Of course, you could
reasonably do it the other way around, however, for a variety of reasons I
think that would be less then optimal.
I agree with all of the above in thoery, but getting people to do this could be hampered by the fact that category feeds would have to duplicate the master feed's metadata in every entry...or at least should...since the WG decided against creating a method of doing it only once in aggregate feeds. That duplication makes me hesitant to embrace this idea wholeheartedly. But I don't know of another solution that the WG hasn't already rejected.
It would be great if we could permit one feed to subordinate itselfShould we add an "authoritative" link type, as recently discussed, and change this wording to work with that?
to another, however, I don't that would really change the fact that there
really should be only one source for an entry.
Agreed on both points.
I'm not sure what you're getting at here. However, it seems to meSearch results feeds: more difficult. If the search criteria don't change, then it's the same feed, so the prior sentence applies.
that a search results feed is almost always going to be an aggregate feed
and thus should be filled with entries that have source elements identifying
other feeds.
But, I suspect that I haven't understood properly what you were
trying to say. Please try rephrasing.
Oops, my mistake--I was thinking of searches of non-entry resources (such as web page searches) being published as a feed. Defining a method for generating ids for such things is a little too far outside of the core of what we're doing to both with. You're right, of course.
Anyway, I've added "originating" three times in that last sentence, making it "An atom:id value that has been used with one entry originating in a particular feed MUST NOT ever be used with a different entry originating in the same feed, and SHOULD NOT ever be used with an entry originating in another feed." Does that address the question sufficiently (within the scope of the Pace at least)?
