Lists Home |
Date Index |
From: "Henry S. Thompson" <firstname.lastname@example.org>
> Eric vDv wrote:
> > Isn't it necessary to introduce a XPointer scheme to identify the (or a)
> > schema(s) which should be used to evaluate the bare names then?
> > ... Although it wouldn't be concise it would be fully "deterministic"!
> We could do that, but it would be wrong (in my view). Wrong because
> it violates locality -- a barename link with name XYZZY is to what the
> _target_ establishes as is its XYZZY ID, not the source. ... The _user_
> does that by setting up the processing environment, in either case.
If I send a request for a fragment of a document, and use some kind
of XPath 2 based on the additions to the information set from schema
processing, then I need to have knowledge of the schema used.
Otherwise I cannot ask for anything by name, because the only
names I can be sure about are the ones in the actual tags. The
schema could use anything.
This comes down to a trust/policy issue: has the remote site/applicaiton
told me what schema is being used, and have they committed to maintaining
the names in their schemas?
So using the PSVI not only tightly couples local processes, but it
tightly couples linked processes too. In the absense of a mechanism by which
an XPath 2 query can tell which schema was used to create the PSVI,
isn't this a bit fragile?
It seems to me that it is very desirable for a link to be able specify exactly
which PSVI it wants. Actually, I guess a "lesser" test is enough: to test
the schema that is used at the remote end can have a common notional
base with the schema at the local end (i.e. the types have the same names,
and one restricts or extends the other, in ways relevant to the query).
The big problem in the idea that we can link using the schema that we want
to use is that the other end is infact probably not schema-processed PSVI
at all: it is probably a TAI interface to some non-XML data. The term
PSVI probably sets expectations a little squiffy.