Re: DOCBOOK: Announce: DocBk XML 3.1.7/XSL Stylesheets V0.18

[Norman Walsh <ndw at nwalsh dot com>]

/ Joerg Wittenberger <Joerg.Wittenberger@pobox.com> was heard to say:
| Norm wrote:
| > #PCDATA and both inline and block content; so the DTD can't
| > prevent you from saying: 
| > 
| >   <entry><para>...</para>foo<para>...</para></entry>
| > 
| > But there's nothing we can do about it, as far as I can see.
| 
| Urgh.  Wouldn't it be better to change the table model here.  We
| already have "entrytable!" and "entry".  How about splitting "entry"
| into <!element entry (#PCDATA|%inline;)+ > (probably the content model
| of a para) and <!element entryblock (...)> (sorry, it's not worth
| right now to look up the correct entities for the content model) with
| the block level elements?

Perhaps, but it would make make the conversion from DocBook SGML
to DocBk XML much less obvious. I think we might very well want
to do something like this for DocBook 5.0.

OTOH, having alternate <entry>'s wouldn't be compatible with the
OASIS Exchange Table Model (or CALS) so I think it would be
better to fix this in the content model of Entry. The easy way
to do this would be to remove "#PCDATA" from the content model
and add phrase. Then you don't have pernicious mixed content, you have:

<!ELEMENT entry ((%inline-things;)+|(%block-things;)+)>

But now all PCDATA in entrys has to be wrapped in <phrase> or
something similar and, well, that's tedious for authors. Which
is why we didn't fix pernicious mixed content back in the
DocBook 2/3 timeframe.

But with XML's requirement that all content models that include
#PCDATA be repeatable, optional OR groups, we probably have to
fix it now.

                                        Cheers,
                                          norm

-- 
Norman Walsh <ndw@nwalsh.com>      | Always keep a song in your
http://www.oasis-open.org/docbook/ | heart--it's like karaoke for the
Member, DocBook Editorial Board    | voices in your head.