OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.


Help: OASIS Mailing Lists Help | MarkMail Help

tm-pubsubj-comment message

[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [Elist Home]

Subject: RE: [tm-pubsubj-comment] Re: PURL and stability of PS Identifiers


I should start this message by saying that I fully agree with your position
on URIs an have been pleasantly surprised at the extent that I have agreed
with what you have said. But I do have a couple of comments on this posting:

>>one PSI <=> one Indicator <=> one Identifier (URI) <=> one subject


>>> A URI is a URI is a URI. If we provide a canonical URI, anybody
>>> should be able to do anything he is used to with a URI, including
>>> setting bookmarks or copying an URL from the browsers URL
>>>field into a mail to a colleague ...
>>Agreed. This is the only way it can make sense.

You can only bookmark URIs that are bound to a supported protocol. Not all
URIs are prepended by their protocol (only URLs are expected to have this
characteristic). So to say that an arbitrary URI can be bookmarked is wrong
(unless it can be bootstrapped to an appropriate resolution service such
bookmarking has no purpose).

>>> May be the coming OASIS member section may provide a
>>> persistent *hosting* (not redirection) of PSI sets. This would
>>> work.
>>As Lars Marius said, I don't follow you at all on that path. It has
>>been clear from the very beginning of the reflection on PSIs that
>>the notion of any central registry is to be ruled out, because it will
>>create an unsustainable bottleneck.
>>At the opposite, the recommendation aims to enable the creation
>>of *distributed* PSI registries. Moreover, there is no reason why
>>OASIS should be more efficient in hosting PSIs than any other
>>organization. If OASIS is able to host stable URIs, why Library of
>>Congress or ANSI or UN would not be able to do so?

URIs are always stable as strings - no problem. What can change is what
happens when the URI is dereferenced or resolved. The stability of
resolution of URIs not a technical issue it is a social one.

>>Either it is possible technically, for an organization serious about
>>it, to maintain a stable set of URIs - if not forever, at least until next
>>week ;-) either it is not a sustainable eventuality, because of the
>>congenital volatility of the Web. Were I convinced that we are indeed
>>in that last case, I would say "Forget it all, PSIs will never work".

However serious an organization is, there is always the possibility that an
unforeseen event will destabilize an infrastructure that does not address
the requirements of longevity and availability. Enron, Andersen, et al were
serious organizations but who knows what will happen to their domains when
the liquidators start their fire-sales. Only by addressing the longevity
problems directly can organizations be confident that resources will be
available when needed.

Fortunately, there is a growing awareness of this problem and initiatives
are starting that address long term persistence of information about
resources. PURL is a low cost mechanism that assumes the owner of the
resource will maintain the resource. DOI [1] is a more complex
infrastructure that can have provision for transferring management
responsibility for identifiers under certain circumstances (the Library of
Congress uses the same technology as the DOI for its internal resource
management [2]).

Both PURL and DOI are about administration of what a URI dereferences to. As
this is a social issue, technology alone won't do. There needs to be
services and policies that allow the administration/maintenance of resources
over time.

>>But I don't figure why maintaining stable URIs should be impossible,
>>without using the convoluted PURL way. Maybe so far, people have not
>>cared enough about it. If we come to the point of considering that having
>>stable PSIs is critical, certainly we'll get them, the same way we got
>>worldwide phone numbers and ISBN when it became critical to have them.

Stably resolving URIs are not impossible - it just needs the social will to
make them happen. But I think two things are being mixed up here. Phone
numbers and ISBNs have the same properties as URIs. They can be created in
error (I can just make up a phone number or ISBN), they can be wrongly
assigned, the thing that they are assigned to can change (yes publishers are
known to reassign ISBNs), etc. So you wont always get what you expect when
you dial a telephone number, or plug an ISBN into Amazon.

So while this group should probably not dictate how URI dereferencing should
be achieved, it should probably identify characteristics that will increase
the likelihood that PSIs can reliably be dereferenced to an intended

[1] http://www.doi.org/
[2] http://www.handle.net/



Eamonn Neylon
Manifest Solutions
John Eccles House, Oxford Science Park
Oxford, OX4 4GP, United Kingdom
T: +44 1865 338033; F: +44 1865 338100

[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [Elist Home]

Powered by eList eXpress LLC