When is data Atom-like?
http://en.wikipedia.org/wiki/Web_syndication
...puts emphasis on the ability to "provide other people with a
summary of the website's recently added content"
A very direct reading of the data format suggests to me that the
questions Atom is most organized to answer are:
What's new?
For each new thing:
... who wrote it, when?
... how do I get the rest of it?
If the questions a user asks are very different from this (e.g. "what
is most popular" or "what is biggest" or "what is this person's email
address" or "when is this user free") then Atom is not organized as
optimally to answer the questions.
Of course this is just looking at 4287, not at extensions that
significantly change what questions atom is organized to answer.
Lisa
On Jun 21, 2007, at 1:37 PM, Tim Bray wrote:
On Jun 21, 2007, at 1:08 PM, Yaron Goland wrote:
A particular question I would really love to hear a concrete
answer to though is - what are the characteristics that make a
data set look like a 'publication'? In other words, what is a
loose check list of things to look for that let you know if the
information you want to expose is a good candidate for ATOM?
I think the answer is: data that's Web-like. I.e. where you have a
universe of resources, each with its own identifier, and which are
expected to use hypertext links embedded in message bodies to refer
to each other. Obviously, the motivating use-case that got the
ball rolling was blog-post authoring, but it's starting to look
like the data space for which the Atom protocol is useful is quite
a bit bigger. -Tim
