Re: [docbook] SGML support

[Pierre Attar <pat@tireme.fr>]

Le 22:58 20/12/2004, vous avez écrit:
>At the DocBook TC meeting this month, I took an action to point out
>that our current plans for DocBook V5.0 will probably have less
>support for SGML than the current DocBook V4.x line.
>
>Our plan of record is to work with a RELAX NG schema from which we
>will generate a W3C XML Schema and an XML DTD. The DTD is not likely
>to have much in the way of parameterization. It's going to be a
>"compiled" product much more than a "source" product. This is a
>consequence of the fact that we're taking advantage of some of the
>expressive power of RELAX NG. The DTD will be significantly "looser"
>than the schema.

Even on the XML side, I se problems on automatic conversion from Relax. 
More precisely, for me, the problem of this approach is that if your 
company does not handles relax, then it becomes difficult to use docbook 
anymore... as soo as you make integration of your own spcific features 
(elements, attributes and content models).

Before taking any decisions, it will be interesting to know :
- how many docbook people are using each XML modelling plateform : dtd or 
schema or relax
- how many people has made customization that requires extensions and 
redefinition.

I recently try to customize the docbook DTD and then generates my W3C 
schema (using trang) in order to be able to integrate the result within a 
corporate set of XML W3C schema definition ... believe me, it was not so 
obvious and there was a lot of thing to suppress in the DTD in order to 
fully define them on the schema. There was also a need to XSLTize the 
resulting schemas in order to change features that where ne so good after 
the trang translation. This was not a trang problem but much more a 
preparation of the integration of now contents models using schemas.

I really prefere the old way (4.2) where enginnering designs where taking 
into account the specificities of each modelling environment (PE for DTDs, 
complex types and redefinitions for schemas, etc.) in order to help people 
that where in a specific modelling environment to use the best features of it.

Imagine: if a company is a DTD or a W3C schema expert and they want to make 
docbook integration, they will need to become Relax expert just for this 
need ! I'm quite sure it is not the best way to think about.

Even more, how will it works with standards like MathML or SVG where, as of 
today, there ar no official Relax standard ? We will need to make post 
trang conversion in order to remove trang generated information for these 
namespaces and replace it by the standard ! That will, once more, 
complexify integration needs.

Here are some though, if you want me to develop more and to provide 
samples, just ask.

Regards, Pierre



Pierre Attar (mailto:pat@tireme.fr)
Consultant en informatique documentaire XML
Consultant in Structured Document engineering

Projet "Mutualiser l'effort de montée en compétences sur XML"
http://www.mutu-xml.org/index.html