2006/6/7, Robert Yates <[EMAIL PROTECTED]>:
Thomas Broyer wrote:
> and that this is needed because entries
> might not always be atomically retrieved (otherwise, "permaLinks"
> would been enough).
I don't understand what you mean "atomically retrieved" and hence why
permaLinks would not have been enough.
By "atomically retrieved", I mean that you don't necessary have an
Atom Entry Document for each entry in a feed (actually, many CMSes
don't currently provide Atom Entry Documents, the only Atom documents
they provide are one or two feeds).
As Atom goes beyond content management à la blogs, there are also use
cases where the is no "alternate" representation: the entry in the
feed is the only representation of the resource, there's no way to
identify it using an URI, other than explicitely associate one through
the atom:id element.
Finally, atom:id is different from a permaLink in the sens that it's
only an identifier that might not be dereferenceable (or, to be more
accurate, that might not return anything but an error when
dereferenced): it's not a link, it's an identifier, which has the form
of an IRI, just as XML Namespaces are identified by URIs.
The advantage of such a design choice is that you can move entries
(hence changing "permaLinks") without changing their identifier. This
is particularly useful when, for example, switching from TypePad to
Blogger, and then to WordPress.Com, and then to a self-hosted blogging
system, and then changing the blogging system (and changing feed and
entry URIs, even if it's not a Good Thing [1]).
A Good Practice for these identifiers –at least when you're an
individual– is to make them "tag" URIs [2]
[1] http://www.w3.org/Provider/Style/URI
[2] http://www.taguri.org/
--
Thomas Broyer
