Zhiwu Xie wrote:
Hi,
Just a question on the syntax of atom:author. Is it the author of the
content an entry refers to, or is it the one who posts the content to a
content management system therefore generates an atom entry? I checked
google blog and it seems the later is the case (snippet below), but then
earlier discussions seems to suggest the former, e.g., multiple
authors/contributors discussion.
There's no right answer.
The same question can be asked on all the other metadata, e.g., rights,
updated, published, etc. A more general question is, what all these
entry metadata are about. The spec says they are all about the feed or
entry, but it's not immediately clear to me what does the entry metadata
are about, the XML element itself or the content an entry points to.
This has much different implications, e.g., if atom:right says do not
copy, does it mean this atom entry cannot be aggregated by other Atom
feed aggregator or the contents cannot be distributed within the
atom:content?
(I assume you mean "implication" in a technical sense)
The only sensible answer I can give is that these facts are typically
about whatever is denoted by atom:id. If your system for whatever reason
needs to disambiguate the referent of metadata, the best you can do is
make sure it's internally consistent, and document that. I would expect
that to result in treating atom:id as a primary key.
The WG chose not to depend on technology that might help with this kind
of issue (such as RDF), and Atom does not have its own semantics. Atom
then doesn't go beyond common sense interpretation - ie, your rights
attribution issue is as likely to be settled as a matter of case law as
specification. It's easy enough to get muddled around whether metadata
is about the entry, the markup you received representing the entry, or
some other entity. Even then formal models are limited and a surprising
number of standards and notional keys have had problems like this
(Dublin Core is a good example for the web, abuse of SSNs as identifiers
is a classic).
There are/have been efforts to create formal models of Atom, dependingon
Semantic Web technology, but none of these have standing as they are not
products of the WG - all Semantic Web vocabularies boil down to URI
ownership so the WG has to agree that certain URIs can stand for Atom;
these do not have be in the Atom/AtomPub namespaces.
cheers
Bill
