CR 1.8 Align dmdSec, techMD, rightsMD, sourceMD and digiprovMD ID attribute requirements with actual practice

(WITHDRAWN after discussion at the METS Editorial Board meeting Dec. 10. The Board felt that the current "required" use restrictions on the ID attributes of the elements in question are not out of line with general practice--at that the referencing of ID values at the dmdSec, techMD, rightsMD, sourceMD and digiprovMD level from DMDID and ADMID IDREFS attributes should be encouraged, as opposed to referencing ID attribute values at lower levels within these elements.)

Currently the METS schema requires the ID attribute on all elements of mdSecType: dmdSec, techMD, rightsMD, sourceMD and digiprovMD. The original assumption was that METS users would want and need to link in all metadata contained in mdSecType elements at the root mdSecType level (dmdSec, etc). And I think that there was an additional assumption that DMDID attributes would always reference ID attribute values on dmdSec elements; and that ADMID attributes would always reference ID attribute values on the techMD, rightsMD, sourceMD and/or digiprovMD elements. This assumption was bolstered by the fact that some xml validator implementations at one time would reject cross-namespace IDREF/ID links. So, for example, they would reject an attempt to reference an ID value on a mods:mods element within a mets:xmlData element from a DMDID attribute on a METS div. This is no longer seems to be the case; and actual practice diverges from our original expectations. Some METS implementations (like Library of Congress I believe)link from DMDID attributes to ID values on the mods:mods element or lower level mods:relatedItem elements as appropriate, rather than to ID attribute values at the dmdSec level. The current restrictions necessitate ID values where they are not needed in some implementations, and suggest a consistency of implementation which in the end the schema is powerless to enforce. Its also difficult to explain the requirements in METS documentation (as I am currently trying to do). Therefore, I recommend that the ID attributes on the mdSecType elements be made optional.

(Added by Rick Beaubien)