Jindrich Marek wrote:

>I know there was some Roma problem discussed here but I don't know if
>this is the case. I tried to use Roma to generate a DTD for fulltexts of
>manuscripts for the Manuscriptorium system
>
>
...

>I got an empty document.
>
>
>
just before you press the final submit, Roma shows you a schema version of
the view of the TEI. If you can bear to repeat the work, and send me the
that chunk of
text, I can find what went wrong. Its likely that someone broke the Java
setup on the server
here

>1. What is exactly different in Roma and Pizza Chef implementation of P4
>(in the DTD structure)?
>
>
they start from different sources, and work in entirely different ways.
Roma takes a Relax NG schema,
and converts it to DTD.  Pizza reads DTD files and mangles them.

>2. Do you recommend using Pizza Chef to generate DTDs?
>
>
only for experimental purposes. the Pizza Chef is the production tool.
Roma is
an experimental replacement working with an alpha version of TEI P5.

Roma should NOT be regarded as supported, reliable, correct, normative etc!

when P5 is released, the Pizza Chef will probably not work with it;  if
it does not, it is unlikely
to be fixed to do so, I think.  However, the Chef will still serve P4
into the medium-term future.

>3. How to convert a Relax NG schema to DTD (is here a simple way which
>is easy to understand)?
>
>

using "trang", from http://www.thaiopensource.com/relaxng/trang.html.
But it will only convert a
subset of RelaxNG schemas. The job of Roma is to massage the schema to
make this work.

Sebastian Rahtz