On 11/14/2017 10:42 AM, Eliot Kimber wrote:
> At the risk of splitting hairs (ok, at the certain risk of splitting
> hairs), I don’t think a schema can be said to “describe how data is
> to be structured”.
Well, I was trying to see how close to the language of Roger's proposed
statements as I could get ...
> Any declarative schema is simply a set of constraints to which
> instances governed by the schema do or don’t conform.
I actually see this statement as being essentially equivalent to mine:
"... and may be used to constrain that structure" - in the sense that
one can use conformation to the schema as a constraint. To me, we are
saying much the same thing, but maybe you still see some split hairs.
> ...So a schema, by itself, can only be an incomplete definition of
constraints, no more.
Yes, and therefore the interest in using different kinds of validation,
like Schematron, to complement an XML Schema.
TomP
