Indeed... I tend to believe that "in the end of the day" only the main
stakeholder of the concerned resource in that moment will be its unique
responsible, so you should consider addressing that since the first day...

As until now we are still not having as part of the Internet/web
infrastructure any generic technique/service do resolve URNs, than that turns
to be simply a quite irrelevant concept (you can use URNs internally to your
services, but it'll be always your internal responsibility to resolve
then...). That is what really what the DOI is: an agreement of a specific
community to address this issue in a common agreed way (but you have to be
member of that community to make use of that...).

In that sense purl.org, tinyurl.com, handle.net, etc. are just more examples
of things that might look useful, but only until they'll be maintained by
those with a direct control and interest on it. If you don't control that,
soon or latter you'll face trouble...

Preventing that, the National Library of Portugal simple registered many years
ago the domain "purl.pt", and has been managing it as its own purl/proxy
service for the National Digital Library. This means that any object in that
service is published with a persistent/friend URL (also a URI, sure...) in the
form "http://purl.pt/xxx" (xxx is simply a "first come first get it" integer
number...)

I took this simple but conscientious decision more than 10 years ago, and in
fact it proved to be VERY EFFECTIVE. It made it possible all those thousand
URLs to persist several changes in the internal technology (and, off course,
even to changes in the institutional DNS of the library itself: ...,
"biblioteca-nacional.pt", "bn.pt", "bnportugal.pt", who know what else will
come...), with a very low maintenance of the service itself (it is only a very
simple proxy service, after all, with a simple back-office that can be easily
used by any external service to manage its own URLs... just like the purl.org
provided by the OCLC...).

Since the Nat. Lib. is a governmental entity, it is expected that even some
possible future "less clever/aware" political/administrative/technical
"decisor" will understand that the purl.pt DNS can easily be maintained
independently of any technical infrastructure or organizational/political
existence or shape... I'm just sad that more public organizations in the
country developing digital libraries, archives, museums, repositories, ... did
not engaged in this simple concept (that was a role for the Ministry of
Culture to push, but... you guess it...).

José Borbinha






-----Original Message-----
From: Repositories discussion list [mailto:[log in to unmask]]
On Behalf Of Andy Powell
Sent: Thursday, October 07, 2010 11:36 AM
To: [log in to unmask]
Subject: Re: Migrating repositories and abandoned URLs

Without wishing to start a religious war, I'd avoid the 'URN' option on the
basis that it's hard to think of an identifier that is less Web-like. <ducks>

As Ian Stuart pointed out in the previous message re Handles:

3) The Web [tm] is based on "cut'n'paste the address line", so even if a
Handle *is* minted, and the Handle Server *is* top notch & spiffing.... 
people won't be using it

The same applies to DOIs. <ducks again>

As I've mentioned here before, a good starting point is to think very hard
about the domain name used for the repository at the outset - like don't embed
organisational structure or the name of the repository software into it - but
I guess it's too late for that in many cases. On that basis,
repository.xxx.ac.uk is much more likely to be persistent than
eprints.lib.xxx.ac.uk.

Cool URIs arise from good planning before any URL is minted at least as much
as from picking up the pieces after things go wrong. <dives under desk>

Andy

--
Andy Powell
Research Programme Director
Eduserv
t: 01225 474319
m: 07989 476710
twitter: @andypowe11
blog: efoundations.typepad.com

www.eduserv.org.uk 

-----Original Message-----
From: Repositories discussion list [mailto:[log in to unmask]]
On Behalf Of Christian Gutknecht
Sent: 06 October 2010 16:09
To: [log in to unmask]
Subject: Re: Migrating repositories and abandoned URLs

For the content of our repository we plan to implement a persistent
identifier. Either DOI (http://www.doi.ethz.ch/index_e.html) or URN:NBN
(http://tinyurl.com/25e7zel), it's not quite clear yet which one we will
choose.

For both system you have to generate the persistent identifier first and make
the identifier visible in the OAI output. The resolver services will then
harvest the identifiers and connect it to the current URL of the repository.
Should there be a repository migration or URL change, you just have to make
public the new base URL to the resolver service, which then will harvest the
identifiers again and will update the identifier with the new URL. This sounds
"really simple" to me. But I've no idea, if it will really work that easy,
when it's needed.

Best regards

Christian Gutknecht


Am 06.10.2010 14:43, schrieb Leslie Carr:
> [Second attempt to send]
>
> When institutions migrate from one repository platform to another, they are
obviously concerned that the metadata from their records transition as
completely as possible.
>
> However, it doesn't always seem to be the case that the "persistent URLs"
that were coined by one repository are kept live by the next repository (or an
adjunct piece of software). The effect is that all the legacy links break when
the old service is switched off.
>
> This seems to be a fundamental issue for services that are supposed to
supply "sustainable access" in an "interoperable" fashion, precisely to allow
maximum flexibility for the users!
>
> Does anyone have any experience of this? What do platform developers need to
do to make the smooth transition of legacy URLs "really simple" as opposed to
"technically possible"?
>
> Les Carr


--
Christian Gutknecht

Main Library University of Zurich
Winterthurerstrasse 190
CH-8057 Zürich
Tel. +41 (0)44 63 54162
www.zora.uzh.ch
www.oai.uzh.ch