How should the Registry handle mirrors?
rplante at ncsa.uiuc.edu
Thu Jun 24 07:36:04 PDT 2004
First off, anyone who is concerned about these issues should bookmark
Clive's and Pierre's posts: Clive's for a good summary of the problem,
and Pierre's for the real world experience.
Although we have discussed this topic extensively, there's definitely no
harm in keeping it in everyone's mind since, as Tony points out, we have
not settled on a solution. Just to catch up those who haven't been in
this discussion, I'll briefly summarize some other bits of the discussion
1. Role of Title, ShortName, and Publisher
On Thu, 24 Jun 2004, Clive Page wrote:
> Is it
> sufficient for the Title element to be identical (so all Vizier clones are
> simply called "Vizier" (or "VizieR"?). The Identifier (URI) will
> obviously be different, but what about the ShortName, and the Publisher?
Tony answered the main question about how to register mirror status. I
just want to add that there no requirements about what the values of Title
and ShortName should be, other than they are chosen by the provider.
Thus, applications should *not* rely on them as unambiguous identifiers;
they are for human consumption. Publisher is a little different because
you can attach an identifier. However, our model is that a mirror of
resource maintained by a different organization is essentially a
re-publishing; that is, the new org is the publisher of the mirror.
2. On Logical Identifiers
An approach from the DL world is to associate a logical identifier
with a set of resources that is location independent. Twice we (I) have
attempted to insert this concept into our XML metadata and both times the
idea was shelved by the WG as needing more work (for the reasons Tony
alluded to). Our current thinking is that we will see how the "mirror-of"
relationship works for now. I think this is a wise approach.
3. Who determines mirror status
The "mirror-of" relationship implies two publishers, and it would be nice
if they both acknowledged this in some way. However, at the moment, the
publisher of the mirror is the one that asserts this claim.
More information about the registry