[osis-core] OSIS_01093

Todd Tillinghast osis-core@bibletechnologieswg.org
Mon, 15 Apr 2002 15:16:56 -0500


> Open Questions that need resolution:
> 
> target attribute on reference: (How can they point to non-biblical
> materials outside the text if they must validate against
referenceType?)
If we want reference to point to things other than those that start with
Bible.* then we should allow that in the regex for referenceType in
OSISScripture.xsd.  We can either enumerate the church fathers or we can
further extend referenceType as defined in OSISScripture.xsd in
OSISChristianTexts.xsd.  Which will nicely OR with the regex in
OSISScripture.xsd.  This of course does not have anything to do with the
regex for milestoneID and milestoneIDREF.  They can be more restrictive.
We can if we want to in OSISScripture.xsd, or some other schema define a
regex that allows an open set of reference systems and names within the
reference portion of a referenceType.  It is my belief that the target
in <reference> should be of type referenceType.

> 
> OSISID/OSISIDREF: what is their function now?

The only benefit is that the door is left open for future options.  That
is not to say that xs:ID itself can't be redefined but that would impact
things like milestoneID.  This way the attribute that is spread broadly
within the schema would have a hook if we wanted to use it without
having to change OSISCore.  (The same argument could be made for "type"
but a weaker argument)

> Verify intent on regex:
> 
> Todd Tillinghast wrote:
> 
> >I made the following modifications to OSIS_0191.
> >
> >Fixed what seem to be errors:
> >1) Added ID and lang to <div>
> >
> Shouldn't ID, lang and type be pulled from global? (Probably error in
> earlier version that did not have <div> with global attributes.) Fixed
> to call global plus special ones.
> 
I think that type is useful several places, but not everywhere.  I don't
have much use for lang myself but I know others do an will deffer to
their judgment.  I think IDs are useful but can really take them or
leave them on most elements.  I think the most useful places are the
same places there are referenceAttributes (<div>, <p>, <lineGroup>,
...), as well as <note>, <figure>, and <title>.
> >
> >2) the target attribute in <reference> is not of type ReferenceType
> >
> If we make target of referenceType, that constrains it to pointing
only
> at Bible verses as validated by referenceType. In other words, I
cannot
> use osisCore to refer to  any of the Church Fathers for example. (They
> have their own reference systems.)
> 
See not above.
> >
> >3) made target in <reference>, <ntProphecy> and <otPassage>
refereceType
> >
> Yes to <ntProphecy> and <otPassage>, no to <reference> see reasons
under
> #2.
> 
> >
> >4) removed the regex from referenceType in OSISCore (because it is
ORed
> with
> >what ever regex is put in a redefined referenceType)
> >
> Sure, good catch!
> 
> >
> >Made the following adjustments that seem be oversights:
> >5) Created an attribute group referenceAttributes and put refStart
and
> >refStop back on <p>, <list>, <lineGroup>, <body>, <div>, <item>, and
> <line>
> >as optional attributes
> >
> 
> OK, these are pointing inside the document (or at least supposedly)?
> 
> >
> >6) Created a refWorkType Type that can also be restricted to conform
to a
> >regex in a redefined instance and made refWork in <text> of type
> refWorkType
> >and made it required.
> >
> OK (I think, let me think about it for another minute or two)
> 
> >
> >7) Created a milestoneID and milestoneIDREF types to be used instead
of
> >OSISID and OSISIDREF in milestones and left OSISID and OSISIDREF the
same
> to
> >be used instead of xs:ID.  My intention is NOT to cause OSISID and
> OSISIDREF
> >to be constrained in the same fashon as referenceType, milestoneID,
and
> >milestoneIDREF but to provide a hook for the future for some other
> purpose.
> >Also changed the names of the attributes in milestones to MID and
MIDREF.
> >
> OK, but what is the function of OSISID and OSISIDREF? Both function as
> ID and IDREF in normal XML, but are not further defined in
osisScripture
> (I assume that was intentional).
> 
See comments above. 
> 
> >
> >
> >Things you probably inteded that I think should be different but left
> ALONE
> >because you (Patrick) want it the way it was.
> >1) <year> is still derived from xs:string!!!!!
> >
> Sorry, error!!! (Now gYear, which allows 1998, etc.)
> 
> >
> >
> >Not sure but these look like errors:
> >1) I'm not sure which elements should have TEIForm attributes and
which
> >should not.  Not all elements have then though.
> >
> Not all elements are the equivalent to TEI elements. Those that are
> (will look again) have the TEI form attribute.
> 
> >
> >2) Why is the TEIForm for <copywright> fixed at "availability"?
> >
> <availability> is an element in TEI that holds copyright information.
> Has an attribute for status (free, restricted, unknown) and a <p> for
> other content.

OK, just checking.

> 
> 
> >
> >Question: Do we need to/want to give OSISCore a suicide pill make it
> >abstract to force encoders to use a derived schema?
> >
> Temping but ultimately probably not a good idea. Hopefully we will be
> responsive enough that we can control the release a new versions with
> added features.
> 
> 
> >
> >Need to not allow the literal "." to the regex in CoreScripture.  Is
this
> >correct?  If not can you comment on the correct syntax.  No time to
> research
> >since the TAX man is still standing outside my door.
> >
> No, the syntax \. allows a literal full stop to appear in the regex.
The
> Bible.Prot is just a string (which includes a full stop).
> 
> If
> 
> ^(.,/.)
> 
> means (All that has gone before) but not followed by full-stop, comma,
> back-slash, or full-stop(again?)
> 
> Don't you want ^(. , / ) (written with too much space to make it
> visible) to exclude full-stop, comma, back-slash?
> 
> Probably better if you could tell me what you are trying to validate
as
> it is not apparent from the regex.
> 
> Bible.(somevalue)..Book.(chapter but optional?).Verse(but optional?)
is
> that the intent?
> 
> Let me know, will try to hack something together and send later this
> afternoon.
> 

Best example is in refWorkType in OSISScripture.  I would want to
preclude Bible.T.J.E as a refWorkType.  As it sits this is possible.
Further we would want to force at least a non literal "." and non blank
space character between the literal "." characters.  

Also want to ensure that ".." only happens as we want it to.

I can give more examples after the tax man gets his share.

> Patrick

Todd