[osis-core] Index Syntax

[Harry Plantinga]

Well, when you start thinking hard about an issue you
come up with all sorts of additional consideration...I'm
going to try to mention some important ones.

1. Index entries in books look like this:
  Subject1
    Subject2  37, 92, 143

I'm going to call the outer entry heading (Steve calls it
key1) and the next heading Subject2 (Steve's key2). I'm going
to refer to the location identifier text as the "title".  In
this example there are three index entries, with three different
titles -- 37, 92, 143.  In print books the title is normally
the page number or some variant thereof -- e.g. 143n2 for p143,
footnote 2.

2.  Most books have only 2 levels of index headers, i.e. subject1
and subject2, but I have seen (old) books with 3 levels.  Some word
processors (e.g. OpenOffice, Word) limit the number of levels to 2.
We'd have to decide how many levels to allow.

3.  Subject1, subject2, and title text that is held in an attribute
can't contain markup.  You couldn't have an entry that says
<i>See</i> p43.

4.  We can't use page numbers for titles anyway -- page numbers
don't make much sense for electronic books. Perhaps we should
use canonical identifiers, e.g. X.iii.5, where the underlying
link is to a particular spot in X.iii.5. However, these can
get kind of ugly and they have lots of internal structure which
is distracting.  Another option would be to use 1, 2, 3, ...
A title attribute could be used to specify the title to use.

5.  If we are willing to go with generated titles (X.iii.5 or 1,2,3)
we don't need a title attribute. If we are willing to have entries
that have no markup, we can use attributes for Subject1, Subject2,
etc.  In that case, we can use Steve's proposed syntax:

<index index-name='xxx' key1='...' key2='..' key3='...'/>

Maybe 3 levels is enough.  I suppose this is a reasonable compromise
between utility and simplicity.

-Harry


> -----Original Message-----
> From: owner-osis-core@bibletechnologieswg.org
> [mailto:owner-osis-core@bibletechnologieswg.org]On Behalf Of Patrick
> Durusau
> Sent: Wednesday, June 26, 2002 1:44 PM
> To: osis-core
> Subject: [osis-core] Index Syntax
>
>
> Harry,
>
> Still working on the current and a proposed revision of the pointer
> syntax (which I think needs a bit more clarity) as well as incorporating
> your observations and remarks.
>
> Since it (may!) be a simpler issue, what about the <index> element syntax?
>
> Steve suggested:
>
> <index index-name='subjects' key1='sacrifices' key2='cows'
> key3='preparation of'>...</index>
>
> To avoid any more overlapping problems than we have already, I would
> suggest:
>
> 1. <index> be an empty element with attributes.
>
> Clarification: does index-name mean the type of index? subject versus
> name versus geographic place? (Assume so but wanted to be certain.) If
> that is the case then each index could hold only one index-name/type
> along with associated keys but if it is an empty element, suppose you
> could have as many as you liked at any point in the text.
>
> Allow index to appear anywhere that milestone can appear?
>
> Comments, suggestions?
>
> Patrick
>
> --
> Patrick Durusau
> Director of Research and Development
> Society of Biblical Literature
> pdurusau@emory.edu
>
>