#155 Define canonical way of referencing TEI element definitions
As an alternative to using a TEI-specific convention for citing URLs, we
might want to consider adopting one or another of the standard systems
for permanent identifiers, for example DOI (Digital Object Identifier).
In the case of DOI, there would be some fairly minimal costs involved to
register the DOIs with one of the national registries, but some more
extensive costs in human time to set up a good system for mapping from
our existing nomenclature to a set of DOI identifiers. (Publishers
typically use DOIs to identify individual books or journal articles, but
they can be more granular: every reference page in the Guidelines could
have its own DOI, for example.)
The advantage of doing this would be that anyone accustomed to using
DOIs for citations could simply cite a DOI. So instead of the reference
for <name> in the Guidelines
http://www.tei-c.org/release/doc/tei-p5-doc/en/html/ref-name.html
we would use something like
doi:10.1111/tei-p5-doc:en:ref-name.html
which would be resolved to the actual Web address by a DOI resolver.
(Many journal DOIs are mostly numeric but the syntax allows more
human-readable names).
If we seriously want to consider an option like this, we should put out
a call to the librarians/metadata gurus in our community for assistance
with implementation.
I agree with Lou that this is an important issue, but partly because of something it would enable that he hasn't mentioned. Currently we're able to reference an element's definition by pointing to its element reference page. This is a convenient shorthand both in pedagogy and discussions about the TEI generally. What the current website doesn't do is preserve all the major 6th-monthly releases of the TEI so that we can point to the TEI as a changing and evolving thing. That is, we can't easily have a meaningfully referenced discussion on the theoretical or practical evolution of the TEI at a meta level because we don't have an easy mechanism to point to the element definitions for one release compared to another. (The reason I say it isn't 'easy' rather than saying we can't do it, is because we _can_ point to sourceforge and individual subversion commits, but there is no easy way of doing so for a set of elements belonging to release 1.0.3 vs 1.1.3 or something since these are done on a commit-by-commit basis.)
My suggestion would be to archive ad infinitum previous 6th-monthly releases on the TEI website with a footer link pointing to the most recent version. If a DOI system were to be adopted, then this version number should obviously be an optional part of it. (with a :current: or similar for the current release at any time).
But in general I think this idea has merit and should be explored with the TEI in Libraries community.
-James
Actually it wasn't my proposal, I copied it from someone else (can't remember who). To take it firther though, I think we need a clearer statement of what needs to be done.
James and David to come up with proposal for doing this, without using DOI.
David and James submitted a proposal http://lister.ei.virginia.edu/~drs2n/uri-proposal.html (July 2009) but this does not seem to have discussed in council or if discussed not minuted. There is a related and more urgent problem to do with @source attribute
I believe this now to be fixed so am closing this ticket.
Although we don't have a DOI we do have persistent canonical URIs for any reference or other TEI Spec.
e.g. http://www.tei-c.org/Vault/P5/1.4.1/xml/tei/odd/Source/Specs/w.xml for the 1.4.1 version of the w element or
http://www.tei-c.org/Vault/P5/1.4.1/doc/tei-p5-doc/en/html/ref-w.html for its reference page.
Although not as compact as a DOI, I believe this answer the underlying problem of canonical ways to refer to things.
However, if this tickets receives more comments arguing we should go ahead with a DOI scheme, it will be reopened for discussion.