[osis-core] New Schema later today!

[Patrick Durusau]

Guys,

Spent three days for a 2 hour meeting on Friday in Nashville. One day of 
travel and mostly downtime was OK but beyond that....! Saw a lot of 
country music sites and other touristy type stuff with Carol. Enjoyed it 
but hard not to fly apart at the seams from lack of actual work being done.

Back at the main console and should have another schema out later today!

Big Changes:

Quotes are back! If you read Steve's issue list notes, it says we are 
dropping <q>. The discussion was the redundancy of <qStart><qEnd> plus 
<q> and not the marking of quotes altogether. Hmmm, maybe a day or two 
away from it has helped? Nah! ;-)

So to be consistent with our use of segmentation, <q></q>,  with 
next/prev attributes returns! Actually works fairly well (I think).

Note that I will be adding a key/keyRef mechanism on names. Names bear 
an optional nameKey which is a defined keyRef (in schema syntax) so we 
can search for names that have essentially public subject identifiers 
(Joseph but only when it is husband of Mary, not guy with the "long coat 
with sleeves" (one of the many offenses of the NRSV translation 
committee)). What you will see will just be an example that validates 
against xs:string unless over-ridden. Allows maximum flexibility for 
users to declare a variety of simple types that can be validated in 
their documents.

Working on the first chapter of Matthew as a sample text. Note that I am 
dropping all the blockQuote stuff from the TEV version. Not real sure 
where they got all of that to begin with, does not appear in the KJV and 
makes little sense in the context of the text. Note that I really see no 
reason for the redudant use of note/notePart that you see in the TEV 
text. Simply enclose multiple references (already marked as separate 
elements) within a single note container. Note type=x-ref and then 
notePart seems like overkill (to me anyway).

Example: <verse id="Matt.1.25"><note><reference 
target="Luke.2.21"/></note>And knew her not ....</verse>

(old version of reference had targetStart/targetEnd, suggest that we 
stay with simply target, the most common case for this edition. Perhaps 
an optional targetEnd for software that can handle it?)

I have marked all the names and will insert notes from the TEV (even 
though I am using the KJV text).

Need some PSI stuff if anyone wants to have a go at that part.

Should finish the key/keyRef and regex stuff (thanks to Steve!) later 
today. Suggesting that we opt for less validation and more on form for 
the base schema. We can suggest a way to do biblical references, global 
"work" attribute, and suggested "Book.Chapter.Verse" but I see no reason 
to force validation on what is admittedly an ad hoc and after the fact 
scheme that varies from one tradition to another. Essentially arguing 
that we can't solve a mapping problem silently in the standard, need a 
mapping table that supports all the variants and that is used by 
software to do the mapping.

Note that XMLSpy refuses to generate a sample file and incorrectly lists 
the possible roots of the schema as osisText, label and transChange. 
Just defective software since it does validate the schema. Has to be one 
or the other.

Well, better get back to the schema so you can see some of these goodies 
later today!

Patrick


-- 
Patrick Durusau
Director of Research and Development
Society of Biblical Literature
pdurusau@emory.edu