[tei-council] Preparation for F2F meeting

[Syd Bauman]

Overall I am skeptical about the utility of a mere validation
service, but not against the idea; and I think a more robust
"checker" is a great idea.

On the validation part of the checker, my responses to some specific
points that have been raised:

MH> - Do we support only RELAX NG, or do we validate against other
MH>   schemas/DTD, and does it make a difference?

RNG only.


MH> - Do we always check Schematron rules, or never, or do we make it
MH>   a switch?
 
In the user interface, Schematron always. (The API should allow for
separation.)


MH>  - Do we offer validation against the custom schemas (Lite, Tite,
MH>    drama etc.) or just against tei_all?

Yes to Lite, Tite, Simple, and perhaps a few others; no to the silly
ones like "Corpus" and "Feature structures". user should also have
choice to validate against the schema specified in the document via
"xml-model" PI, or against a schema specified.


LB> But wouldn't it be easier to teach such people to use the right
LB> tools for the job?

No, not easier; somewhat harder. But still a lot better. :-)


I think the validation portion of this service should perform
XInclude processing and strip out constructs from namespaces not
declared in the schema (yes, will requires reading the schema unless
it's one we provide like Lite) before validating.

If this service provided really good error messages and an API, it
might be reasonable to use it in front of TAPAS, rather than
re-inventing the wheel.