louburnard wrote: > According to the edw89 draft the proposed mechanism > * should be optional, ignorable, and over-ridable > * should not be TEI-specific > > If the first is a serious requirement, I cannot see any point in the > proposal at all.
The same is true for xsi:schemaLocation and to some extent for !DOCTYPE.
The difference between proposed <?schema?> PI and
!DOCTYPE/xsi:schemaLocation is that <?schema is general and works with
any schema language. Moreover it doesn't rely on legacy stuff (!DOCTYPE)
and it doesn't force you to modify real content of document
(elements/attributes/text nodes). PI allows attachment of more then one
schema.
So if we could agree on fact that there is a need for attaching schema
to document then <?schema is superior to mechanisms that are already in
wide use.
> Given that the proposal has already been discussed and failed to
> persuade one expert forum (the rng users group), why resurrect it now?
That's your perception. My perception is that there are serious
requirements coming at least from from DocBook, TEI and XHTML which
can't be addressed with current standards for schema association. This
argument was silently ignored by people who were against this proposal,
they haven't offered any solution to the problem of having several
schemas for a single namespace.
Instead they persisted on idea of specifying schema externally. Since
this discussion passed NVDL standard was introduced and implemented. But
even NVDL (of which I'm big fan) doesn't solve this problem.
So there are several possible ways to proceed:
1. Simply ignore requirements coming from real-life problems
2. Extend NVDL to cover those requirements (but this would be quite big
change to NVDL, which I do not think happen at least in a short term).
3. Reach agreement on <?schema?>. All users who need this functionality
will benefit from a single syntax supported by growing plate of tools.
4. Do not reach agreement on schema. Users of TEI will have
<?tei-schema, users of DocBook will have <?docbook-schema, .... Both
tools vendors and users will be disgusted.
To me it seems that if we do not want to hide our heads in sand (1.), we
should do 3. and in a long term we can try to make 2. happen.
> For the record, I think this is a waste of time. The most I think we
> should do in TEI P5 is to add a few sentences to SG (introductory
> chapter on XML) explaining how a schema might be associated with an
> instance, possibly giving the rationale for not doing it. And I'm not
> even sure about that.
OK, how do you then will differentiate between TEI and TEI-lite and how
do you will associate RELAX NG schema to TEI instance? Just for the
record. ;-D
Jirka
--
------------------------------------------------------------------
Jirka Kosek e-mail: [EMAIL PROTECTED] http://xmlguru.cz
------------------------------------------------------------------
Professional XML consulting and training services
DocBook customization, custom XSLT/XSL-FO document processing
------------------------------------------------------------------
OASIS DocBook TC member, W3C Invited Expert, ISO/JTC1/SC34 member
------------------------------------------------------------------
Want to speak at XML Prague 2007 => http://xmlprague.cz/cfp.html
signature.asc
Description: OpenPGP digital signature
