James Holderness wrote:
Erik Wilde wrote:
good point. i have not hey thought about how aggregation of aggregated
feeds would work. i guess that for the second aggregator, the fact
that it is aggregating an aggregated feed should not matter. but then
the question is how to treat the <atom:source> and more specifically
its via:via="" attribute. so i guess aggregation of aggregated feeds
does need some special treatment. how about retaining all the
<atom:source> children, but replacing the via:via="" attribute's value
with the ID of the aggregated feed?
There should be one and only one atom:source element per entry,
reflecting the original source of the entry. If you're aggregating an
entry that already has an atom:source, that atom:source element should
be left unchanged.
See section 4.1.2 of RFC4287: "atom:entry elements MUST NOT contain more
than one atom:source element." And 4.2.11: "...by adding an atom:source
child element, if it is not already present in the entry".
thanks for pointing that out. i was (almost) proposing to leave
atom:source unchanged, with the exception of the foreign markup of the
via:via attribute. to not violate the rules you are citing, i could move
this information to an extension element as a child of the entry,
instead of an attribute of atom:source. so the revised version of the
markup would look like this:
<feed xmlns="http://www.w3.org/2005/Atom"; xml:via="...">
<link xml:id="feed1" rel="via" href="http://example.com/feed-A"; />
<link xml:id="feed2" rel="via" href="http://example.com/feed-B"; />
...
<entry>
<via:via ref="feed1"/>
<source>
<link rel="self" href="http://example.com/feed-A"; />
in the first-level aggregation, and the second aggregator would create
something like this:
<feed xmlns="http://www.w3.org/2005/Atom"; xml:via="...">
<link xml:id="id1" rel="via" href="http://example.com/feed-C"; />
<link xml:id="id2" rel="via" href="http://example.com/feed-D"; />
...
<entry>
<via:via ref="id1"/>
<source>
<link rel="self" href="http://example.com/feed-A"; />
which would make clear that (a) the entry has been originally published
in feed-A, but has been aggregated through feed-C ("id1"). this pattern
could be repeated through any number of aggregators. or would the
via:via element no simply be a repetition of <link> and it should be
like this:
<feed xmlns="http://www.w3.org/2005/Atom"; xml:via="...">
<link xml:id="id1" rel="via" href="http://example.com/feed-C"; />
<link xml:id="id2" rel="via" href="http://example.com/feed-D"; />
...
<entry>
<link rel="via" href="http://example.com/feed-C"/>
<source>
<link rel="self" href="http://example.com/feed-A"; />
if i use this design, the only new markup i need is xml:id on the feed's
li...@rel = 'via'] children, but in the end my application would have
more strictly defined rules on how to handle <link> when aggregating
than what is defined by atom itself. the rules would be:
1. if you aggregate a feed, you must list that feed as a <link
rel="via"/> and you must add an @xml:id for it. this id must be stable
over time.
2. for each aggregated entry, you must add the feed from which you
aggregated it as a <link rel="via"/> on the entry.
these rules ensure that whenever i read an entry from an aggregated
feed, there will be a feed/li...@rel = 'via'][exists(@xml:id)] for the
entry/li...@rel = 'via'].
cheers,
dret.
