[tei-council] Re: [rng-users] time to really allow instance to name schema?

Sebastian Rahtz sebastian.rahtz at oucs.ox.ac.uk
Mon Feb 19 12:58:16 EST 2007


Jirka Kosek wrote:
>> you could embed a stanza of NVDL inside the document?
>>     
>
> But this would modify "information content" of document itself
it seems to me that this is what the proponents of the
idea want to do, they genuinely want to enhance
the document.
> - you will have to ignore NVDL elements during XSLT processing
>   
Yes, of course; I don't see that as any big deal; just like I
ignore most metadata elements during daily processing.
Sometimes I want them. Sometimes I also want to consult
the schema, so I will read the NVDL.

And of course it did not stop XSD from adding
their horrible attribute....

The days when you got the "real" content of a document
by stripping out all tags is surely past?
> - how would you implement validation in a streaming mode if you must
> lookahead NVDL snippet first
>   
doesn't the same apply to a PI? it has to be inside the root
element, so you have to open the document. and it applies
to the XSD element.
> I think that for situations where schema can't be specified indirectly
> using NVDL or similar technology using <?schema?> PI is the best
> solution -- easy to use, easy to implement, no side-effect in existing
> toolchains.
>   
but it does require a new ad hoc agreement, and one which cannot
be checked against a schema, or contain any useful documentation.
I agree it would work (if tool vendors wanted), but I don't think
its as much fun as it could.

Anyway, I tend to agree with Lou that I want different
schemas at different times, and that I'd rather not be
dictated to by the document...

-- 
Sebastian Rahtz      

Information Manager, Oxford University Computing Services
13 Banbury Road, Oxford OX2 6NN. Phone +44 1865 283431

OSS Watch: JISC Open Source Advisory Service
http://www.oss-watch.ac.uk




More information about the tei-council mailing list