[osis-core] Overlapping Summary??

[Steve DeRose]

At 06:45 AM -0400 06/21/02, Patrick Durusau wrote:
>Greetings,
>
>Note that I have qualified this post with ??? as I am not sure the 
>following represents a consensus on this issue. It is an attempt to 
>state what may be a consensus with a request for comments and 
>corrections.
>
>For OSIS 1.1 (our current release candidate):
>
>Overlap will be treated by segmentation of (primarily verses but 
>other elements as well) the crossing element. (Need to specify which 
>one gives way.)
>
>Segments of verses that cross will bear the same verseID (it is not 
>type = "xs:ID" so duplicates don't matter) along with next/prev 
>attributes that have the string value of the verseID followed by an 
>underscore and a lower case ASCII letter (a-z) to indicate its 
>position in the segment.
>
>(Reasoning that XQuery/XPath will find all segments bearing the same 
>verseID attribute and return the entire verse.)
>
>Questionable summary: I have heard both that we should do a separate 
>module with empty elements (bearing element name semantics) as well 
>as using my earlier suggestion of the default XML empty elements 
>(where content model does not require content).

My impression was that we were leaving the empty-element solution out 
of 1.1 entirely, and putting it off to 2.0 (unless there are 
particular exceptions, quote seeming like the best candidate, as 
Harry described). Does that make things any easier short term?

>
>The difference being that for OSIS 2.0, that the semantics of the 
>empty elements be declared to have a container relationship, while 
>in OSIS 1.1, they are simply empty elements.
>
>I actually like the Steve/Harry suggestion on defining the semantics 
>rather than my more back-door hack on the default syntax of XML. I 
>think I could make my solution work as is, but it might be more 
>confusing in the long run and that is a definite disadvantage.
>
>Suggestion: Should I clean up the current schema to focus solely on 
>segmentation (with appropriate attributes) and work on an additional 
>module that has the most common crossing elements with the semantics 
>of milestones? Same attributes that they have now in fuller form 
>(although we need to proof those as well). The module could make it 
>clear that these carry the semantics of being empty elements but 
>that they carry attribute information (with proper use) that will 
>allow that semantic to change to containership in OSIS 2.0?

I'm reluctant on that, because I think we need more time to work out 
just what we'll need on those empties to make them behave, and I'd 
like us to introduce a general solution that will cover a lot of 
ground.

>
>Suggestions for syntax or commentary?
>
>Comments?
>
>Patrick
>
>--
>Patrick Durusau
>Director of Research and Development
>Society of Biblical Literature
>pdurusau@emory.edu


-- 

Steve DeRose -- http://www.stg.brown.edu/~sjd
Chair, Bible Technologies Group -- http://www.bibletechnologies.net
Email: sderose@speakeasy.net
Backup email: sderose@mac.com, sjd@stg.brown.edu