[osis-core] open issues

[Steve DeRose]

At 02:49 PM -0400 06/14/02, Patrick Durusau wrote:
>It looks like I will be out most of tomorrow but back at full tilt 
>by either late tomorrow or Sunday. I need to take Carol to the 
>American Library Association meeting and was planning on going to 
>the exhibit hall. ;-)
>
>Troy,
>
>So far I see at least four ways of dealing with the problems you 
>have noted with bible markup (and I do agree it is a serious problem 
>and we do need to make recommendations about best practices):
>
>1. Segmentation (problematic in some cases as you have pointed out)
>
>2. Elements can be written with content or milestones (if my 
>analysis of the XML standard is correct and I think you originally 
>suggested this one.)

This is possible syntactically, but a convention that an empty 
version of the X element (whatever), counts as a starting or ending 
milestone for a crossing version of X, seems a bit arcane. Seems like 
it would be clearer to either have everything permit X, Xstart, and 
XEnd variants (kind of gross), or to at least have anything being 
used in empty form as pseudo-milestones, be required to indicate that 
somehow (say, a whichEnd attribute or something).

But I think my inclination for now is to keep it simpler, even though 
it does get gross for some gross cases like that triple-decker 
quotation. We can put that example into the doc specifically, to show 
the worst case; I imagine there aren't a whole lot that get that 
ugly....

for later things like annotation, at word level there's no problem 
with crossing, and at higher levels I suspect most annotation will 
live out-of-line and point in anyway. Thoughts from Kirk?

>
>3. Use the milestone element we presently have with a type attribute

In that case we could make a convention that if it's to be treated 
like something else, you have to put that something else into the 
type attribute.

>
>4. Todd's suggestion of creating types of elements with verseID 
>(assuming I understood it properly) which looks like an elegant way 
>that I had not previously considered.

This is kind of cool; I have mixed feelings on it. It makes 'verse' 
really a marginal phenomenon -- wherever possible, verseIDs would 
appear on other units like p, line, etc; verse itself would only show 
up when there isn't any other unit that does it. In that case, does a 
verse consisting of two lines get tagged  on both lines (like a 
segmented verse), or as a container around the 2 lines?

>
>Agreeing that we have to address the problem, comments or 
>suggestions on what would constitute best practices? (I don't think 
>that we can dictate for all cases how markup should be done. Well, 
>we can but then the standard becomes less useful as situations arise 
>that we cannot now forsee.)
>
>Patrick
>
>
>
>Troy A. Griffitts wrote:
>
>>Not sure if anyone knows or cares, but most all of my issues 
>>brought up in my last messages are still on the table:
>>
>>archive, if anyone needs the old messages:
>>
>>http://www.bibletechnologieswg.org/cgi-bin/lwgate/OSIS-CORE%40BIBLETECHNOLOGIESWG.ORG/archives/
>>
>>Not wanting to be a wrench in the works, but I still feel our gears 
>>are turning fast down a wrong path.
>>
>>Specifically: segmentation isn't elegant and we can't anticipate 
>>all overlapping cases.  Try a few verses from a Prophetic text 
>>(e.g. Ez 39:1-) e.g (not from Ez).
>>
>>1. Then God said, "Say to my people: 'Thus says the LORD, "Repent! 
>>2. You backsliding markup authors."
>>     How often have I desired to give you a good markup. 3. But you 
>>were unwilling.
>>
>>
>>Try to add phrase level annotation, critical markup and other tags 
>>to that and you have no hope of a human understanding or getting it 
>>right, muchless our schema to allow such things.  Anything can be 
>>forced to work.  I just don't see why our milestone approach wasn't 
>>about 3000x more elegant.
>>
>>Or even my unanswered suggestion of a having all tags allowed to be 
>>used 2 ways with a global 'milestone' attribute (kindof like Todd 
>>is suggesting with roles, but not exactly).
>>
>>global attributes mStart and mEnd
>>
>>That would allow 2 types of markup.  All tags to either be used as:
>>
>>XML Container:
>><verse ID="Gen.1.1">In the beginning</verse>
>>
>>or
>>
>>XML Milestone / Logical Container
>><verse ID="Gen.1.1" mStart="uniqueID1" />In the beginning<verse 
>>mEnd="uniqueID1" />
>>
>>
>
>--
>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