On Aug 30, 2005, at 3:08 PM, Walter Underwood wrote:
That is one kind of order. Other kinds are relevance to a search term
(A9 OpenSearch), editorial importance (BBC News feeds), or datetime of
original publication (nearly all blog feeds, not the same as last
update).
Well put. It's obvious that there exist feeds that have ordering
requirements that are orthogonal to atom:updated.
It's plausible, but not obvious, that the first two kinds of order
Walter enumerates are instances of a class which we might label
"importance" or "priority". It's also plausible, but even less
obvious, that there are going to be lots of places where non-updated
order matters in a way that has to do with "importance", to the
extent that a generally-implemented extension will be useful. It's
also plausible, but not obvious, that there's no usefully-
generalizable notion of "importance", that the semantics of this
stuff are really application-specific, and that interested parties
should go and do their own NetFlix or A9 or BBC extensions. Or just
cram your Netflix queue in a single entry.
Fortunately, the format spec, with its "assigns no significance"
language, doesn't get in the way of people who want to work on this
problem. I can see two generalized approaches; the first is a
singleton attribute of child of <atom:feed> that says "in this feed,
<atom:entry> order reflects importance". The second is such an
attribute that says "in this feed, an ordering by the following
namespaced-element child of <atom:entry> reflects importance". I
prefer the first. I haven't looked at the Microsoft proposal.
Clearly, the WG lacks consensus, at least in the short term. -Tim
