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

 


Help: OASIS Mailing Lists Help | MarkMail Help

oasis-member-discuss message

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


Subject: RE: [oasis-member-discuss] Comments on Artifact ID requirements


Although I just returned from an extended vacation (and must admit that I
have not had time to review the Artifact ID requirements in depth yet),
nonetheless the points Peter makes in this message are very good ones, and
in fact are central to XRI (Extensible Resource Identifiers) specifications
developed by the XRI TC (http://www.oasis-open.org/committees/xri). XRIs
extend (and are backwards compatible with) URI/IRI syntax and add features
designed explicitly for abstract identification of resources -- and
resolution into concrete representations of those resources -- following the
precise guidelines Peter describes.

A further benefit is that XRIs standardize several of the key types of
identifier metadata Peter enumerates (such as version, language, date -- see
the XRI Metadata Specification V2 Committee Draft 01,
http://www.oasis-open.org/committees/download.php/11854/xri-metadata-V2.0-cd
-01.pdf). 

Lastly, the XRI 2.0 specifications include full support for HTTP proxy
resolution so all XRI syntax and resolution capabilities can be deployed
immediately using available HTTP infrastructure.

A suggestion, then, is to arrange a meeting between the Artifact ID team and
the XRI TC to discuss the requirements and see if XRI 2.0 can help OASIS
solve its resource identification needs.

Mary, Jamie - who should we contact about this?

=Drummond 
================
Drummond Reed
Co-Chair, OASIS XRI & XDI TCs
http://public.xdi.org/=Drummond.Reed 


-----Original Message-----
From: Peter F Brown [mailto:peter@justbrown.net] 
Sent: Tuesday, August 02, 2005 4:53 AM
To: oasis-member-discuss@lists.oasis-open.org
Subject: [oasis-member-discuss] Comments on Artifact ID requirements

Dear all:

Whilst I appreciate that much effort and time has gone in to this draft
recommendation, I am very uncomfortable with the proposals as they stand. I
have only just had an opportunity to look the document through and compose
some comments: possibly too little and too late.

The TAB asks that comments and proposals be specific and concrete, so here
goes:

Proposal:
Withdraw the working draft; rethink requirements for identification and
addressing as distinct issues.

Reason:
There is a fundamental conceptual flaw in the proposals, and one which we
unfortunately see all too often on the web, and is perpetuated by a similar
conceptual flaw in the W3C's own thinking on the issue.

This flaw is to confuse the identity and the address of an artifact,
ultimately concluding that they are the same. The consequence of this flaw
would be that the proposals, if applied, would irretrievably confuse and
combine two important concepts that require to be considered as distinct in
any coherent and lasting information architecture.

Argument:
There is little disagreement that it is useful to identify any given
artifact. Giving it identity allows for the artifact to be cited, referenced
and talked about in a manner such that there is no ambiguity as to its
nature.
Ambiguity can enter the picture, however, from the moment that an artifact
is considered in a particular context: which version of the artifact? Which
format? In which language? In sum, which "representation" [1]? The reality
is that every combination of such factors (language, format, version, etc.)
leads to a specific and unique artifact: sometimes it is useful to identify
and reference such specific representations; other times it is more useful
to reference and identify the "abstract" entity.

The objective of the proposed requirements is essentailly to be able to
reference, cite and link to only representations and not abstract entities,
but there are two steps to such an approach which are unfortunately rolled
into a single issue in the proposal:
- firstly it is necessary to identify an artifact: to be sure that we are
talking about the same "thing";
- secondly, it is necessary to dereference that identity *in context*.
It is this second step that is often so unstable because too many contextual
issues are left implicit in the process of resolution (naming > addressing)
whereas they should be made explicit.
This is implicit in the proposal as it calls on TCs to even make redirection
a specific conformance requirement to send a user from an "abstract" URI
representing a particular artifact to a "concrete" network endpoint resource
(or "representation").

There is currently no generally accepted distinction between a conceptual
"entity" and a "real" artifact which, I beleive, is not only useful but
essential to disambiguate the issues related to identification and
representation addressing.*

The structure of the proposed identifier explicitly admits that semantics
can be deduced from the ID. These semantics - such as "version", "stage",
etc. refer to aspects of a work's process lifecycle and not to inherent and
non-dissociable properties of the work itself.

The identifier structure also implies a hierarchical relationship between
components of the identifier, whereas the relationship is purely
associative.

Some examples:
I can talk about a "work" (for example a TC's charter, an information model,
a specification) as well as representations of that work (the charter as
first published; a revision to the charter;version 5 of a particular schema;
the French language translation of the current version of a specification).
The current proposal makes no distinction between the conceptual, abstract
entity and its representations.

A document does not "have" a "version": version is a property of a
relationship between an abstract entity and a representation, it is not an
inherent property of the representation itself. The representation does not
change (hence, consistent with the principle of permanent URLs) but only the
nature of its relationships through associations.

Ultimately, any persistent referencing architecture should only use the
identity of an abstract entity which should be de-referenced dynamically and
contextually to then point and direct to a specific concrete entity. One of
the greatest sources of "broken link" problems is precisely the lack of such
information architectural modeling.

Any attempt to establish an addressing model that relies upon and is based
on context-dependent links will ultimately fail and is not sustainable. It
is far more stable and dependable to build artifact identity based on the
conceptual layer.


[1] "Representation" - a specific rendition of a given "abstract" or
conceptual artifact


Peter F Brown
-------------
Chair, CEN eGovernment Focus Group
-------------
ICT-Strategy Unit
Austrian Federal Chancellery
-------------
Co-Editor, OASIS SOA Reference Model
-------------





---------------------------------------------------------------------
To unsubscribe from this mail list, you must leave the OASIS TC that
generates this mail.  You may a link to this group and all your TCs in OASIS
at:
https://www.oasis-open.org/apps/org/workgroup/portal/my_workgroups.php 





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