The XML Exchange Table Model specification doesn't actually appear to
have any language about conformance--the strongest thing their is is a
caution that making content models less restrictive might impair
interchange.
However, it's hard to see how making tgroups optional could do that,
especially in the DITA case, since a table with no tgroups isn't a
useful table anyway (except in the context of conref, which would be
resolved before a table processor would be expected to see it).
I would also suggest that any OASIS-table-aware editor that depended on
tgroup being required was just poorly designed.
But it's not that big a deal--it just seemed sort of odd to me and was
one of the few places within body content that a likely conref target
cannot be empty (I suppose all of the lists have the same problem--you'd
have to have one empty list item).
I also agree that the current conref= approach is just not right--it
reflects an SGML legacy that was a poorly-thought-out feature in SGML.
I would much prefer a replacement for conref= that was based on XInclude
and that allowed, for example, the name of the referencing element and
the referenced element to be different but still have a clear
declaration of what the allowed reference targets are (e.g., some form
of "reftype" attribute as we provided in the HyTime standard).
But reforming conref= is definitely a 2.0 thing, as it needs to be done
in concert with a more general reformation of DITA's addressing and
linking semantics.
So I withdraw this feature request.
Cheers,
Eliot
----
W. Eliot Kimber
Professional Services
Innodata Isogen
8500 N. Mopac, Suite 402
Austin, TX 78759
(214) 954-5198
ekimber@innodata-isogen.com
www.innodata-isogen.com