* Daniel Aleksandersen <[EMAIL PROTECTED]> [2008-02-01 20:30]:
> 
> On 2008-02-01 15:41, Aristotle Pagaltzis wrote:
> > * Daniel Aleksandersen <[EMAIL PROTECTED]> [2008-02-01 05:15]:
> > > It is not globally unique on a per my feed basis; but it is
> > > globally unique on a per…me…basis.
> >
> > Put the ISSN into a tag: URI instead of using the urn:issn:
> > scheme.
> 
> What good would that do?

Tag URIs contain your domain, so they provide a globally unique
identifier, even if you’re using your ISSNs in them, which are
not by themselves globally unique.

* Daniel Aleksandersen <[EMAIL PROTECTED]> [2008-02-01 22:30]:
> The question was regarding archive feeds. Should these use the
> same ID as the main feed (issn), or should I append an archive
> date range as a second identifier (issn#date)?

The entire point of atom:id is that entries are entities in their
own right and may appear in multiple different feeds; so the
atom:id for a particular entry should always be the same no
matter which feed that entry is in. If you use different IDs for
entries in your archive feed, you are effectively asserting that
the entries in the archive feed are not the same 

The feed might even be provided by a search engine such as
Icerocket, Bloglines or Google Blog Search; hence the requirement
for atom:id to be *globally* unique, not just unique within a
feed or even your domain. So urn:issn: URIs are not usable as
atom:id values. If you use them and some other publication uses
them in *their* feeds as well, then your entries won’t be
distinguishable by atom:id from the entries of that other
publication when a user goes to a feed search engine and gets
entries from both your and the other publication in the result
feed.

Another issue is that atom:id values are not supposed to be used
for any purpose other than identifying an entry, and so are not
supposed to be normalised in any way – only compared character-
by-character.

So in other words, don’t try to put anything meaningful in
atom:id – just make sure it is globally unique and never changes.

Hence why I said to use tag: URIs.

Regards,
-- 
Aristotle Pagaltzis // <http://plasmasturm.org/>

Reply via email to