[osis-core] Summary on: osisCore_Candiate_1.1_003 - 10 reference to an entire work.

[Patrick Durusau]

Guys,

I have simply copied the relevant posts for this thread below, more for 
my benefit than yours, ;-), and lets see if I can isolate the issues:

1. Todd wants to have a reference to an entire work, but wants the 
string to include a specification of the reference system?

I am implying that from Todd's

<reference osisRef="Bible.KJV(Bible.TEV)"/>

Otherwise, I would just say use:

<reference osisRef="Bible.KJV"/>

2. Steve and Harry appear to agree that we need a single string that 
identifies:

1. reference system

2. work

3. osisID

Is that a correct summary?

I  think I have played a large role in the confusion over osisWork and 
its syntax because what I was hearing was apparently not what was intended.

When I hear osisWork, I take that to be a reference to an attribute in 
the header that  defines that work. A common example that we use is 
Bible.KJV. To answer Harry's question from another post, no, I don't 
think that necessarily has significance outside of the document because 
we have not defined a canonical set of identifiers for Bibles (or other 
works).

However, having said that, I tend to confuse the osisWork attribute 
Bible.KJV with the reference system that is also known as Bible.KJV (not 
arguing for this, trying to make explicit the source of my confusion so 
don't bother correcting my misimpression, just watch for it should it 
reappear) ;-)

So, when I see, Bible.KJV:Gen.1.1, I am assuming the Gen.1.1 is a valid 
location identifier within the work that was specified. Probably a bad 
assumption but I suppose I was thinking that it would make little sense 
for me to use an improper reference system for that work, for example:

<reference 
osisRef="Bible.KJV:Title_26.Subtitle_A.Chapter_1.Subchapter_C.Part_III.Subpart_A.Sec_351"/>

//

(don't ask, you really don't want to know, it is a real citation however)

Now, my confusion does lead to the unfortunate case that Harry lacks a 
way to specify a separate reference system for a work within a single 
string. It also could lead to confusion between encoders who choose 
versions that have different reference systems for the same text (KJV?) 
but who use the same Bible.KJV as the osisWork value.

So, I assume from the lack of a choruses of YES! and Harry's post 
indicating that the reference system, work and osisID need to be in a 
single string, that my notion of placing the reference system 
information in the header is dead? (Perhaps all for the best, a brief 
surge of sorrow, moment of silence ...., moving on)

Suggestion:

Let's take the evening to look at the various syntax options for the 
osisWork, osisID, osisRef complex (while I try to wade through all the 
other posts) and see if we can fashion a solution that avoids my 
implicit reference system notion and is not too ugly in a regex.

With a little luck, you should have a new version (with hopefully 
improved internal documentation by late tomorrow afternoon). Sorry for 
the delay in getting another one out but I would prefer to fix the major 
issues so texts can be produced to test the solutions.

Patrick

Todd:

Todd Tillinghast wrote:

>Currenly we have a nice mechanism to point to a reference within a work
>but not to an entire work or reference system.
>
><reference osisRef="Bible.KJV(Bible.TEV)"/> is invalid.
>
>
>But <reference osisRef="Bible.TEV:Gen"/> is valid.
>
>If we change the regex to no have a ":" then we have no way to
>distinquish between a reference to an entire work or reference system
>and a reference that is defaulting the reference system.  We could allow
>the following:
>
><reference osisRef="Bible.KJV(Bible.TEV):"/>
>
>One ugly thing is that the reference system is not needed when referring
>to the entire work.
>
>Do we need a <workReference> element?  It is tempting to add another
>attribute to <reference> named osisWorkRef but then both attributes
>would be optional.  I would rather see another element and have the
>attributes be required.
>
>Todd
>
Steve:

Replies that it is tacky but later clarifies that is it tacky that the 
proposed reference is invalid

Todd:

Replying before Steve corrects on the "tacky" (technical term in W3C 
debates ;-0 )

>
>With out this sort of syntax we have no way of optionally saying that I
>mean this reference identifier in this reference system in this
>translation.  This is a necessary behavior.  There have been alternate
>syntaxes thrown around over the past year but this seems to make the
>most sense.
>
Steve:

Agrees that use of aliases should be deferred for now.

> Right. I'm sorry I wasn't clear, I meant that the fact that your 
> example is currently *invalid* is tacky -- not that the example is. I 
> think we *should* allow reference to the whole work, and I don't mind 
> the slight ugliness of the dangling colon below, etc. 


Harry says:

>
>
>I would like to see a single string that identifies reference 
>system, work, and osisID.
>
>Todd's proposal is fine with me, or any syntax that puts it all
>into a single string. The workID should be optional.
>
>One issue that seems to me to be unresolved is this: does the 
>workID merely point to a <work workID="xxx"> element, or does
>it have some external significance?  If external, how are these
>things defined, kept unique, etc?  Same thing for reference
>system identifiers.  
>


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