Lists Home |
Date Index |
From: "Mike Champion" <firstname.lastname@example.org>
> It's unfortunate that the debate has been cast in terms of
> "characters" vs "types." It should be about "which types are so
> universal that they can be supported deep in XML without undermining
> the reasons for its success" and "how can type information be passed
> around in a way that is useful for type-aware applications and
> unobtrusive to character-only applications."
I think also it is a divide between people who do symbolic processing and
people who do general data processing. (Of course it is not a total divide,
but it is a continuum seemingly with few people in the middle.)
Publishing and semantic users generally only need structures, text and symbols:
in particular different kinds of names (local names, universal names, system identifiers,
public identifiers, catalogs, RDDL, scoped unique identifiers.) Even
numbers are typically used more for structural constraints (tables column counts).
The end-application can provide much more detailed and semantic-specific
error messages than can be gleaned from the document, often. Especially
errors of integration rather than errors of an individual document per se.