[tei-council] ODD notes

[Syd Bauman]

> 1. Can an ODD document contain more than one <schemaSpec> element?
> The schema permits this, but the current toolkit expects to produce
> only one set of outputs. So either multiple schemaSpec elements
> should generate an error or the current toolkit needs enhancing

I pointed this out to Sebastian a few years ago; he said that the
current toolkit was not about to be able to handle multiple
<schemaSpec>s, so we added the doucmentation into the --help message
of `roma` that says only the first is processed.

I think ODD should allow multiple <schemaSpec>s, but `roma` (and
Roma) should be more forceful about giving an "I'm sorry, I can't
do that" message than it is now. If no one else wants to do this, I
will try to get to it by the weekend.


> 3. If ODD is meant to be fully general there ought to be some way
> of specifying the base collection of modules which are to be
> combined by reference from within a schemaSpec, defaulting to
> "tei". A @base attribute on the schemaSpec might do this (not to be
> confused with xml:base however)

I dont' understand this at all. Can you explain with an example, maybe?


> 4. When deleting (eg) an elementSpec, why is it necessary to specify the 
> module? this is redundant, surely.

Seems silly to me, too. (It would make sense if we could have a <foo>
in module 'bar' and a different <foo> in module 'blort', but we
can't, so it's redundant.)


> 5. Should the text of the TEI Guidelines not include a <schemaSpec> 
> somewhere, generating (presumably) tei_all?

Errr ... it takes us away from our "*all* uses of TEI are
customizations" stand, but other than that sort of principled reason,
I don't know why not.


> 6. We don't specify anywhere in what circumstances order is significant 
> for the declarations within a <schemaSpec>, nor that only the 
> declarations contained by one (either directly or via moduleRef, 
> specGrpRef etc) are used in the output schema. Should declarations not 
> so contained generate an error?

I didn't think order was significant, at least not in one's
customization ODD. I also didn't realize any *Spec declarations were
valid outside of a <schemaSpec>. Are they?

I may not be thinking clearly ... may have a better perspective after
I eat; also will look at #7 then.