[osis-core] Schema issues

[Steve DeRose]

At 10:59 AM -0400 06/11/02, Harry Plantinga wrote:
>I managed to get XMetaL Pro to read (and validate) a document
>with the OSIS core (test 10) schema.

Cool, congratulations!

>
>--------------- duplicate attributes ----------------
>
>XMetaL didn't like the schema at first because some elements
>had repeated attributes with the same name. I found four
>instances:
>   - osisText has "work" attribute and globalAttributes,
>     which also includes "work"
>   - blockQuote has outReferenceAttributes and globalAttributes
>   - div has inReferenceAttributes and globalAttributes
>   - milestone has inReferenceAttributes and globalAttributes
>
>After these duplications were deleted, XMetaL was happy with the
>schema.

Patrick, can you repair these?

>
>------------------ "ews" attribute ------------------
>
>I could figure out what element and attributes were intended to
>be from the name.  The one that entirely mystified me was "ews".
>What is that for?  Could a more meaningful name be used?

ews.... I don't remember either, which adds to the argument for a 
clearer name. We also need some doc here -- got to get that tech 
writer ABS hired rolling.

>
>---------------- format attribute ----------------
>
>In some header elements, there could be more than one format
>for a given element, as well as more than one type.  For example,
>there may be several dates of interest for a document: the date
>it was originally written, the publication date of the print edition,
>the publication date of the electronic edition, etc. These dates
>may be represented in different formats, such as 11 June 2002,
>2002-06-11 (ISO format), etc. I could see using "type" to specify
>which date it is, but what about the date format?

I'd favor having a date-format attribute on there. The TEI has 
'calendar' for BC vs. AD, etc, and 'value' for 'the date in a 
standard form', but no way to specify the particular standard form. 
They also distinguish dateRange, such as for the year range example 
you give below.

Thoughts? My thinking is that the date in content can be whatever the 
author/publisher had, and we should normalize onto a value attribute. 
Perhaps just decide it's always ISO we normalize to? though there is 
an argument for providing Hebrew and Islamic dates, which do not map 
nicely to ISO form.... If we want to provide for those too, it seems 
we should either define a date-format prefix for the value attribute 
(either iso:date='...' or date='iso:...') or someplace else to 
specify.

>
>It may be desirable to have names in two formats: short and
>canonical, e.g. John Calvin and Calvin, John (1509-1564).  Would
>the recommendation be to use "type" to distinguish them? It seems
>more like a "format" to me.

I'd like to leave that to formatting if possible. Though I think we 
should consider adding the TEI dateRange.

>
>Unless there's only one legal format for a header element, I'd
>suggest adding a "format" attribute.
>
>------------------------ question ----------------------------
>
>Where is the reference scheme documented?
>
>-Harry


-- 

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