[tei-council] Re: module dependency

[James Cummings]

Sebastian Rahtz wrote:

>> I agree. The @depends on <elementSpec> would
>> be redundant, because one can analyze the content
>> model.


That makes sense, yes, just analysing the content model gives you the same thing
really doesn't it, and can exploit the class system.


>> However, as I say, what then? If we go back to James'
>> situation with the current schemas, in which he
>> asked for msdesc but not namesdates, the end
>> result is a schema with dangling references. He knows
>> it's invalid, his parser told him so.  So Roma should
>> have done something about it for him. A system which
>> lets you make mistakes and then just crows at you
>> saying "ha ha you fell into the trap, you lose" seems
>> pretty old-fashioned to me.


Yes, that is why I initially complained about it.  It seems as though Roma
should *not* allow me to generate invalid schemas.  It would be nice if the
content models also made this the case, but there are other ways I can use Roma
to make invalid schemas (if I really really try).  Roma should warn me about
when I'm being stupid.


>> I want to write the content model for <msIdentifier>
>> in such a way as to make it _much_ harder to generate
>> invalid schemas. If we just want to report errors, lets
>> forget the class system and modules, and just write
>> content models which say exactly what we want.


-James