[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

It's about the Entries, Stupid!



Walter Underwood wrote:

> If a feed is treated as a single document, supporting per-entry

> xml:lang will require some interesting changes.

I think that there is much, much too much attention being paid to "feeds" in discussions of Atom. Feeds are not the interesting part of Atom…

 

      It's about the Entries, Stupid!

 

      Treating a feed as "a single document" is clearly wrong…

      A feed is a logical, abstract container for a set of one or more entries that have been and might be published. An instance of a concrete XML “<feed>” is simply a reification, potentially partial, of the abstract feed. A reified, concrete, feed is a snapshot of a subset of the total contents of the feed over time. While any instance of a concrete feed will have a finite and knowable number of entries, the abstract feed may contain entries that were omitted from the concrete feed. Additionally, new entries may be inserted into the abstract feed in the future. Thus, the best model for a feed is something like a stream, not a document. While a document, or a version of a document, is something that has fixed content, a feed, like a stream, is or can be constantly changing.

In Atom, the only things that correspond to "documents" are the entries, not the feeds. A search engine should treat a feed as nothing more than a source of entries. Indexing “Feeds” is conceptually the same as indexing a “web server” rather then its web pages. This isn’t just some abstract, conceptual nonsense, it actually relates to the origins of RSS as a format for “Rich Site Summary”. The intent was to produce a set of summaries that stood as proxies or replacements for the actual web pages or parts of web pages on a site. Just as the web pages themselves have distinct identity, so should the rss:items or atom:entries that are or will be used as alternatives to them. What is truly bizarre to me is that even though entries are such clearly important resources, we don’t have a clear mechanism for giving them URI’s and retrieving them as individual objects.

Search engines should *not* index feeds. They should index the entries that are contained in the feeds.

      It's about the Entries, Stupid!

 

            bob wyman