![](https://secure.gravatar.com/avatar/7698ca8517c6b0d840c3eb065bceca3d.jpg?s=120&d=mm&r=g)
On Fri, Nov 08, 2002 at 10:22:56AM +0000, Henning Hraban Ramm wrote:
Hello again!
JM's proposal of FAQ format is only for one FAQ entry. The following one may contain the whole thing: several "recipes" (FAQ entries), each in more than one language.
Your DTD has a number of problems. I have skipped most of this discussion, so I do not know what you want. But the content models are very vague: ANY for question and answer, and it seems you assume ANY for code as well. If you do not specify this more precisely, too much will be possible and nobody knows what to expect. A content model like %text;, declared as (#PCDATA|p|code)+, would be better. Schemas have some extra facilities here, but otherwise I am not sure why they would be preferable. The id in %person should be CDATA, as already remarked. Note that this assumes a tightly organized set of documents. You say nothing about the author or editor, so you assume that the id resolves into an entry in another document, and that the renderer knows which document so that it can generate a hyperlink. You might also use XLink language: xlink:href="author#hraban", which already indicates that the author can be found in the document author in the same directory, with the ID hraban. OTOH, this is less flexible w.r.t. the organization of the document set. Note that Docbook has provisions for QandA: qandaset, question, answer. I think docbook is a prime example of too wide a DTD. But it is useful to try and devise a DTD that is a subset of Docbook. It would make your documents processable with generic Docbook tools. Regards, Simon -- Simon Pepping email: spepping@scaprea.hobby.nl