You'll want to specify the "-c" switch on the RulesBuilder command
The problem is in some sense in the TEI DTDs, in that they
reference GIs in content models that are not defined by a
<!ELEMENT... of their own. But this isn't the TEI's "problem", in
that it is legal SGML. The problem is that poor A/E understandably
doesn't really know what to do with such an element, and thus would
prefer you didn't do that. So RB issues a warning, but supplies a
switch to let you supress the warnings in case you really want to use
an element that you have not defined, whether A/E will know what to
do with the element or not.
What A/E does, by the way, is allow you to insert the element in the
appropriate places, but does not allow the element to have any
content at all (neither data nor other elements). It looks sort of
like an EMPTY element, but isn't -- it's exported with both a start
and an end tag. On the other hand, it seems to be treated like an
empty element on import, in that if such an element has content, A/E
generates an error message at the end-tag. I think this message may
be the result of A/E assuming the PCDATA implies an omitted end-tag;
thus when it gets to the real end tag, it's an error.
If you are using a Mac, btw, I'd be happy to send you binary .rls
files for the TEI base tag sets that I've compiled so far. I think
I've only done prose and verse, but maybe drama, too.
-- Syd Bauman, Programmer/Analyst
Brown University Women Writers Project
[log in to unmask]
P.S. My projection of why A/E does what it does is my own flight into
fancy that may or may not remotely resemble SoftQuad's reasoning.
Not only do I in no way represent SoftQuad, I'm not even an expert
in A/E and RB, I'm just a user.