[tei-council]DTDs (was conformance draft)

[Lou Burnard]

We are not talking about dropping support for DTDs any time soon. We are 
talking about how much of the inner workings of the TEI DTDs should be 
exposed in P5.

In P3 and preceding, DTD technology was moderately new as far as many of 
its users were concerned, and so the Guidelines go to great lengths to 
explain everything about how it works, exactly. There was no pizza chef, 
let alone Roma, so the only way of customizing the thing was to really 
grok those parameter entities.

We have now introduced ODD which is widely regarded as a great step 
forward precisely because it conceals that machinery from the user and 
allows us to specify in the Guidelines how schemas should be constructed 
without going into details about how they should be instantiated as 
RelaxNG or DTD or whatever... sort of like we don't talk about the 
stylesheets needed to format TEI elements.

However, I think there is justice in the Mulberry view that says we 
really ought not to require that everyone use our own ODD engine, and 
that no-one will ever be able to build another ODD engine if we don't 
document how such an engine is supposed to work somewhere.  Note that I 
am not saying "we need a user manual for Roma" (we have one more or 
less); I am saying "we need a specification for how the various bits of 
ODD are actually used to generate Relaxng and DTD schemas"

I am now coming round to the view that this should be a new section of 
chapter 28. I think the DTD part of it would be fairly easy to compile  
(l;argely from all the bits I've been hacking out of other parts of that 
chapter) but I will need help drafting the RelaxNG bits. Something for 
Sebastian to be thinking about on his boat?

Lou


 James Cummings wrote:
> Sebastian Rahtz wrote:
>   
>> Syd Bauman wrote:
>>     
>>> Right. If you wanna use a DTD, the only one you get to use is either
>>> a) the one generated from your ODD via roma, or
>>> b) one you create from the Relax NG schema you stitched together, if
>>>    we go that route.
>>>   
>>>       
>> pretty radical. no DTD modules to be created at all any more?
>>
>> I had the impression on TEI-L that the "DTD bad, schema good"
>> war is not proceeding very well.
>>
>> I'd wager that at least 1/3 of the Council members
>> use DTDs for their day to day work with the TEI...
>> (including you, Syd?)
>>     
>
> I don't see that this matters?  Surely these days those DTDs are generated
> from ODD via Roma?  I haven't used a DTD for absolutely ages which wasn't
> generated from Roma.  Since Syd above in point a) says that DTDs generated
> from roma are OK, then I don't see this as part of the "DTD bad, Schema
> good" debate.  People are still free to use DTDs, they just have to be
> generated from Roma.
>
>   
>>> Furthermore, if you wanna use a DTD we should make it very clear and
>>> abundantly explicit that you are not getting a lot of the validation
>>> that you would get if you were using Relax NG.
>>>       
>> that's a separate matter (though I agree)
>>     
>
> Agreed (though not in Conformance, but Modification section).
>
> -James
>
>