indexterms in footnotes [was: DocBook Technical Committee Meeting Agenda: 18 November 2009]

["Grosso, Paul" <pgrosso@ptc.com>]

As I said, I'm okay with allowing indexterm as a child of footnote
if we feel it is worthwhile making that change (that has not been
the case throughout DB 4.x, so I question the need).

But you're inventing some kind of semantic that doesn't exist.

Unless you have indexterm-start and indexterm-end elements that 
apply the index semantic to all the intervening content or some
other way to indicate to what content an indexterm applies, indexterms 
don't apply to anything.  They are point-wise things.  The point
on the composed page corresponding to where the indexterm element 
occurred is the point that determines the page number in the index, 
and the contents of the indexterm gives the index entry content.

To take an example having nothing to do with footnotes, suppose
you have a indexterm immediately following a para start tag.  If
you claim that such an indexterm applies to the para, then what
happens when that para flows over a page break?  Do you expect
both pages to be reflected in the index?  If the indexterm applied
to the para, you should get both page numbers in the index, but
I claim you should only get the first page number because the
indexterm applies just to the point at which it occurred.

Going back to the footnote, if you claim the indexterm applies
to the footnote, would you expect the footnote callout (e.g., 
the superscripted number) to be a target of the index entry 
as well as the footnote body?

paul

> -----Original Message-----
> From: Rowland, Larry [mailto:larry.rowland@hp.com]
> Sent: Thursday, 2009 November 19 9:49
> To: Grosso, Paul; docbook-tc@lists.oasis-open.org;
docbook@lists.oasis-
> open.org
> Subject: RE: [docbook] RE: [docbook-tc] DocBook Technical Committee
> Meeting Agenda: 18 November 2009
> 
> To me, an index term is at least as likely to apply to an entire
> footnote as it is to an element contained within the footnote.  I
would
> favor allowing it as a child of footnote.
> 
> Larry Rowland
> 
> -----Original Message-----
> From: Grosso, Paul [mailto:pgrosso@ptc.com]
> Sent: Thursday, November 19, 2009 8:34 AM
> To: docbook-tc@lists.oasis-open.org; docbook@lists.oasis-open.org
> Subject: RE: [docbook] RE: [docbook-tc] DocBook Technical Committee
> Meeting Agenda: 18 November 2009
> 
> [I'm not sure Mike's email made it to the lists, so I'm forwarding
> it along with a response of mine.]
> 
> > -----Original Message-----
> > From: maxwell [mailto:maxwell@umiacs.umd.edu]
> > Sent: Wednesday, 2009 November 18 21:32
> > To: Grosso, Paul
> > Cc: docbook-tc@lists.oasis-open.org; docbook@lists.oasis-open.org
> > Subject: Re: [docbook] RE: [docbook-tc] DocBook Technical Committee
> > Meeting Agenda: 18 November 2009
> >
> > On Wed, 18 Nov 2009 15:09:39 -0500, "Grosso, Paul" <pgrosso@ptc.com>
> > wrote:
> > > ...
> > >>       2821653  indexterms in footnotes
> > > ...
> > > First, I note that indexterm is already allowed within paras
> > > within footnotes (at least in DocBook 4.x) which is where I'd
> > > usually expect an indexterm to be.  The discussion during the
> > > TC call seemed not to realize this, so I wonder if I'm confused
> > > or the discussion was confused.  (Perhaps this is different in
> > > DocBook 5.0--I didn't check.)
> > > ...
> > > Given that footnotes are allowed as descendants of footnote
> > > but just not as immediate children, and given that I don't
> > > see also allowing indexterm as an immediate child of footnote
> > > as useful, I suggest there is no reason to accept this RFE.
> >
> > From the DB 5 documentation for <footnote>
> > (http://www.docbook.org/tdg5/en/html/footnote.html):
> >    indexterm must not occur in the descendants of footnote
> > So this is exactly the opposite of what you're describing above
> > for DB4: according to the DB5 documentation, <indexterm>s are
> > allowed as immediate children of <footnote>, but *not* as
> > children of descendents of <footnote>.  Hence the RFE.
> >
> >    Mike Maxwell
> 
> Just to be clear, I did not review--and was not talking
> about--any documentation (for DB4 or DB5) nor the DB5
> schemas.  I merely checked (and tested within Arbortext)
> the XML DTD-based DocBook 4.x doctype in which
> (1) %footnote.mix; (the content model for footnote)
>     does not allow indexterm, but
> (2) does allow para and other elements whose content models
>     do allow indexterm.
> In fact, I note that, in the SGML version, there are some
> exclusions on footnote, but indexterm is not excluded.
> 
> I leave it to the rest of the DocBook TC to decide exactly
> what to do, but my suggestion would be:
> 
> a.  ensure the DB5 schema matches the DB4 DTD in this case
>     so that indexterm is not excluded from children of
>     footnote that would ordinarily allow indexterm if
>     they were not within a footnote;
> 
> b.  ensure the content model for footnote does not allow
>     indexterm as an immediate child of footnote (though
>     I could go either way, but I see no reason to change
>     the status quo here);
> 
> c.  ensure the documentation reflects the schema and puts
>     no further constraints on how indexterm can be used
>     within footnote.
> 
> As an aside, if the DB5 schema does not exclude footnote
> within footnote (the DB4 XML DTD does not [because it
> could not given what one can do in an XML DTD]), then
> that semantic exclusion should be indicated in the
> documentation.  (I did not look at the DB5 schema, so
> this exclusion may already be covered.)
> 
> paul