[osis-core] osisWork, refSystem - Summary

[Todd Tillinghast]

> Guys,
> 
> So I take it the common agreement is as follows:
> 
> 1. Should have attributes on osisText that are osisWork and refSystem.

Yes, I think that they should both be required.  Since we are defaulting
references we must require refSystem.  I am concerned about a required
osisWork for the cases where the osisText element does not represent the
entire work but only a portion (just the book of Bible.TEV:Matt or just
Bible.TEV:Matt.1) would we represent the work is osisWork="Bible.TEV"
which seems to imply that it is the entire work or
osisWork="Bible.TEV.Matt" and osisWork="Bible.TEV.Matt.1" OR be happy
with osisWork="Bible.TEV" knowing that it means all OR part of the work.

> 
> 2. The values of osisWork and refSystem are declared in separate
> container elements, <works> and <refSystems> respectively, by <work>
and
> <refSystem> elements, each of which has an attribute that acts as an
> identifier for that entry.
> 

Yes.

> 3. Q: Do we need an alias attribute on <work> and <refSystem> so that:
>  Bible.KJV.1611 could be used as KJV? (Harry, that is my understanding
> of your requirement, may be entirely incorrect. Please correct if I
have
> missed the point.) Note that the alias would still have to meet the
> requirements of the regular regexes.

Is Bible.KJV.1611 intended to be a global and standardized identifier
for that version or is its meaning only local to the document.  It seems
to me that there is benefit in having it be a global, standardized
identifier.  

IF Bible.KJV.1611 is a global identifier then we cannot simply shorten
its name to KJV in the local document.  That being the case if we must
have KVJ then we must support aliases (KJV=Bible.KJV.1611).  I think
that we should defer this sort of behavior since we can already default
reference systems.

> 
> 4. On the issue of ID/IDREF for osisID, osisRef and osisWork. Note
that
> use of ID/IDREF forces us to obey the XML name constraint, which we do
> not enforce, deliberately, for osisID for example. I will be posting a
> greatly reduced schema to hopefully indicate how we could do
key/keyref
> and avoid that problem.
> 
Agreed.

> 5. <works> should be required in header and be required to have at
least
> one child <work> which contains all the header stuff.
> 


We must also require at least one reference system since we must require
the declaration of a default reference system.

> Comments?
> 
> Will push the next release to 6:00 PM today so I can reach all the
other
> posts that are pending and perhaps have some comments back on proposed
> changes.
> 
> Patrick
> 
> --
> Patrick Durusau
> Director of Research and Development
> Society of Biblical Literature
> pdurusau@emory.edu
>