[osis-core] Are we artificially creating numerous work specific reference systems based on presentation tradition and for the sake of a single, unique OSIS ID for each verse.

[Todd Tillinghast]

We must separate the issue of being able to mark verse starting and
ending points precisely in the text using milestones from that of
self-identifying verse elements (which implies that the identifiers
assigned to the verse element identify a verse or verses(s) that start
and end at the boundaries of the element).

What we have created with the current schema is mechanism that
identifies the starting and ending of verses only as precisely as the
boundaries of the <verse> elements.  If the precision of the verse
boundaries is not sufficient then milestone elements can optionally be
employed.

With that in mind, it seems that in many cases because a translator
homogenizes several verses into a single text block and thus a single
<verse> element that a new "work" specific reference system is NOT
needed but rather that a "standard" one is still in operation.  The
difference being that these <verse> elements are identified by several
verse identifiers rather than a single one.  By forcing a "work"
specific reference system when none was intended by the translators so
that our markup works out nicely with a unique identifier for each
<verse> element introduce additional complexity into the mapping problem
(that will have to be solved later) only to provide UNIQUE identifiers.


That is not to say that we don't need unique identifier as an option to
facilitate elements that are segmented, but osisIDs don't really work
for any case but the <verse> element anyway.  So why not use a
consistent pattern among segmentable elements and not force osisID to be
a unique SINGLE identifier, when I can think of no purpose other than
for segmentation to force osisID to be a single unique identifier.  If
you I have missed a huge angle why an osisID must be a single unique
identifier please correct my flawed thinking.

In summary, if we allow osisID to be a set of one or more osisIDs as we
currently think of them then we will NOT force the creation AND use of
numerous work-specific reference systems AND will strongly facilitate
"interoperability" (for lack of a better term) between documents encoded
using the OSIS schema.  This is not attempting to solve the mapping
problem but only trying to not artificially complicate the mapping
problem for the sake of a single unique identifier.

In most cases that have been brought up thus far, the encoded work is
not really trying to create a new reference system or even a branch off
of an existing one.  Rather the translators assign blocks of text to
either multiple identifiers from a "standard" reference system or assign
a single identifier to more than one <verse> element, post-pending an
"a" or "b" to differentiate the two parts. 

Just because the TEV prints "6b-11" for the block of text in
Matt.1.6a-Matt.1.11 does not mean that they intend to create a new
reference system where the verse identifier is "6b-11" rather it seems
to me that this is the best presentation of the set of verses that the
block of text represents based on the set of verse identifiers from the
"standard" reference system that the block of text the element contains.

This does not suppose that there are no works that truly employ a unique
reference system nor does this suppose that there is not a need for a
mechanism for mapping between reference systems. 

Todd