fo:table-body issue

I am unclear on some parts of our XSL FO table model.
Specifically, I cannot find information in the spec
on the semantics of table-body, in particular, what
it means to have multiple table-body elements.

Our content model for table is:

(table-column*,table-header?,table-footer?,table-body+)

which means there is only one set of table-column specs,
at most one header, and at most one footer, but then
multiple table-body's.  So, I assume that the column specs
and header and/or footer hold for all table-body elements.
So, then, what is the point of multiple table-body's?

Compare the CALS model which is basically:

table
	tgroup+
		colspec*, thead?, tfoot?, tbody

(where each of thead, tfoot, and tbody contain row+).
In this model, tbody is merely a wrapper for all
(non-head, non-foot) rows within the current tgroup,
so there must be exactly one tbody.  And each of
possibly multiple tgroups scopes possibly different
thead, tfoot, and colspecs to the tbody it contains.

I observe that the FO model only allows one colspec
set and head/foot per table, and while the CALS model
allows for more here, I could live with the restricted
FO model.  But then, I don't see why the FO model allows
table-body+ instead of just table-body (that is, exactly
one table-body), and if we do plan to continue to allow
the +, then we need to explain what, if anything, is
the difference between, say, a table-body containing
four rows and two consecutive table-body's each 
containing two rows.

I see the possible alternatives (in descending order
of preference to me) as follows:

1.  make the FO table model more like CALS, to wit:
    table
        table-group+
            (table-column*,table-header?,table-footer?,table-body)

2.  change the FO table model to require exactly one table-body:
    table
        (table-column*,table-header?,table-footer?,table-body)

3.  leave the FO table model as is and explain that the table-body
    FO really has no semantics or formatting effect, it is just a
    wrapper.

paul

p.s.  Is the HTML for the lastest version of the draft
available on the Web?  If so, where?  If not, Anders,
can you email me a zip of it.  (The ps/pdf is not
useful to me.)  thanks.

Received on Thursday, 2 November 2000 12:30:08 UTC