[tei-council] should an ODD say which version of the TEI it applies to?

[James Cummings]

Sebastian Rahtz wrote:
> James raised an interesting question with me about an ODD spec. Should it/can
> it say which version of the TEI it applies to? He points out that given an 
> ODD, you know which elements you dont have (cos you excluded them), but not
> which ones you *do* have (cos that depends on what one gathers from a module
> at compile time). This makes an ODD + instance document incomplete from an
> archival point of view.

Yes, my point being that schema languages come and go (or at least change) and
that the archival format for one's project documentation and schema should be
ODD.  And if it is preserved in ODD, then it should contain all the information
needed to generate that schema.  So either this should mean it records what
stable version of the TEI it was generated with respect to (and the TEI
maintains each of those stable versions and possibly Roma honours requests to
generate schemas against them) or there is an archival version of the ODD which
lists not only all elements you don't have, but also  (perhaps in mode change
but with identical data?) all the elements you *do* have?

> When we reach TEI 5.1, should an ODD be able to say "Get my stuff from TEI
> 5.0, please"?

Yes.  I'd argue that you should be able to do this with any stable milestone
release of the TEI.

> Can anyone propose a notation for how this could work?

Several people have suggested that the TEI needs a mechanism in the teiHeader
somewhere for indicating what ODD/schema/etc. that a particular document
instance is meant to validate against.  If this contained not only a URI of some
sort pointing to the ODD/schema/etc. but a version number, then whatever this
notation is could also be used in an ODD to record what version of the TEI the
ODD refers to.

> or should we say that preservation means ODD + generated schema?

I think that goes against the principle of ODD as One Document Does-it-all, no?
   And it isn't even that really, it is ODD + TEI Database + generated schema,
since the generated schema may not be a full and complete record of what the ODD
requests via the TEI Database. (I'm thinking here of things not being able to be
expressed in XSD or DTD?) I would like to see an ODD which contained all the
information necessary.  If you are presented with an ODD and a generated schema,
then you have the input saying what it doesn't want, and one possible form of
output saying what is there, but not one document saying everything.  While, of
course, I would encourage someone depositing texts with us *also* to include the
generated schemas, it would be nice if there was an ODD which had all the
information in it.   I realise I'm splitting hairs here to some degree.

-James
-- 
Dr James Cummings, Oxford Text Archive, University of Oxford
James dot Cummings at oucs dot ox dot ac dot uk